Zürich, Switzerland Sidst tjekket Oct 12, 2026

ETH Zurich (Swiss Federal Institute of Technology Zurich)

Uafhængig · Åben adgang · Ikke en officiel universitetsside

Hvad vil du gerne kontrollere?

Søg universiteter

Må jeg bruge AI til mit kursusarbejde?Underlagt de gældende betingelser Skal jeg oplyse om brug af AI?Krav til oplysning og undtagelser Må jeg uploade følsomme data?Anvendelsesspecifikke begrænsninger Hvilke værktøjer har jeg adgang til?Underlagt de gældende betingelser

Brug i ChatGPT ↗Inden du indsenderRåd og forholdsregler Opdateringer af poster ↓

Gennemse den aktuelle post efter emne

AI-anvendelse i opgaver og bedømmelser

Underlagt de gældende betingelser

ETH Zurich slår til lyd for en proaktiv og ansvarlig tilgang til generativ AI i opgaveforløb. Studerende bærer fortsat det fulde ansvar for indholdet af ethvert arbejde, de afleverer. Generativ AI kan spille en supplerende rolle (såsom idéudvikling, strukturering eller kodehjælp), men må aldrig erstatte den personlige indsats. Undervisere har myndighed til at bestemme, om og hvordan generativ AI må anvendes i deres kurser.

Officiel kilde ↗

Eksamen

Anvendelsesspecifikke begrænsninger

Præstationsvurderinger skal gennemføres selvstændigt og personligt af den studerende. Brug af uautoriserede hjælpemidler, herunder generative AI-værktøjer, som eksaminator ikke har tilladt, udgør akademisk snyd og kan medføre disciplinære foranstaltninger i henhold til ETH Zürichs bekendtgørelse om disciplinære foranstaltninger.

Officiel kilde ↗

Oplysning om brug

Krav til oplysning og undtagelser

ETH Zürichs originalitetserklæring giver to muligheder: at bekræfte, at ingen generativ AI er blevet brugt, eller at bekræfte, at godkendt generativ AI er blevet brugt, og at angivelsen er aftalt med vejlederen. Bachelor- og masterafhandlinger kræver en underskrevet originalitetserklæring, mens krav til andre skriftlige opgaver afhænger af de gældende studieregler. Ph.d.-kandidater erklærer selvstændigt forfatterskab gennem tilmeldingsprocessen til doktorgradseksamen. Den tilladte brug, formatet for angivelse og eventuelle undtagelser skal følge kravene fra den relevante vejleder og enhed. ETH Zürichs undervisningsretningslinjer instruerer studerende i at være gennemsigtige om deres brug af generativ AI og i at angive, hvilke værktøjer der bidrog til hvilke dele af arbejdet, i overensstemmelse med gældende instruktioner fra underviseren og citationskonventioner. Ved videnskabelig skrivning og afhandlinger bør studerende på forhånd aftale med deres vejledere, hvilken AI-brug der er godkendt, og hvilke angivelseskrav der gælder. ETH Library-vejledningen anerkender en undtagelse, hvor en vejleder udtrykkeligt har accepteret, at en bestemt AI-brug ikke skal angives særskilt. Uden en sådan aftale kan ikke-angivet brug udgøre en overtrædelse af akademisk integritet.

Officiel kilde ↗

Privatliv og følsomme data

Anvendelsesspecifikke begrænsninger

Upload af strengt fortrolige oplysninger til eksterne cloud-AI-tjenester er forbudt. Fortrolige oplysninger må kun behandles i specifikt evaluerede konfigurationer (såsom M365 Copilot Work-tilstand uden websøgning). M365 Copilot Chat er godkendt til interne data. Google AI-værktøjer er endnu ikke blevet vurderet af CISO (vurdering OPEN) og er ikke frigivet til ETH fortrolige, strengt fortrolige eller beskyttede personoplysninger. Ophavsret og databeskyttelse skal til enhver tid overholdes.

Officiel kilde ↗

Institutionelle værktøjer

Underlagt de gældende betingelser

ETH Zürich anbefaler Microsoft Copilot, Google Gemini og Google NotebookLM til undervisningsformål. Chatversionerne i basisudgaven er inkluderet i institutionelle abonnementer på Microsoft 365 og Google Workspace, gratis for studerende og medarbejdere, og de tilbyder virksomhedsdatabeskyttelse, hvor brugerdata ikke bruges til modeltræning. Google-værktøjerne har endnu ikke gennemført CISO-vurderingen (ÅBEN), så produkttilgængelighed udgør ikke en godkendelse af dataklassificering. Værktøjets tilgængelighed betyder ikke tilladelse til bedømmelse; undervisere afgør tilladelser til bedømmelse.

Officiel kilde ↗

Forskning og publicering

Underlagt de gældende betingelser

Forskere og studerende er fuldt ud ansvarlige for integriteten af deres videnskabelige arbejde og skal verificere AI-genereret output, referencer og fakta. Generative AI-værktøjer kan ikke angives som forfattere eller citeres som autoritative kilder. Specialer kræver underskrevne tro og love-erklæringer; ph.d.-kandidater erklærer uafhængigt forfatterskab ved tilmelding til ph.d.-forsvaret. Forskere skal overholde gældende retningslinjer fra bevillingsgivere (f.eks. SNSF) og forlag.

Officiel kilde ↗

Flere gennemgåede emner

Akademisk integritet

Anvendelsesspecifikke begrænsninger

ETH Library stiller Turnitin Originality til rådighed for berettigede professorer og lektorer til kontrol af tekstlighed, underlagt dataklassificering og vejledning om ansvarlig, selektiv brug. Kun offentlige eller interne dokumenter må uploades. Lighedsscreening er adskilt fra detektering af AI-forfatterskab; ETH Zurich advarer om, at teknisk detektering af AI-genereret indhold er upålidelig og ikke bør behandles som afgørende bevis. Overtrædelser af GenAI-retningslinjer såsom brug af uautoriserede hjælpemidler eller manglende oplysning om deres brug er underlagt disciplinære sanktioner i henhold til eksisterende regler for præstationsvurdering og tro- og loveerklæringen om originalitet.

Officiel kilde ↗

Undervisningsvejledning

Vejledning med opgavespecifikke krav

Undervisere opfordres til at tilpasse undervisnings- og bedømmelsesstrategier til GenAI, drøfte acceptabel brug med de studerende, tydeligt kommunikere, hvilke tilladelser der gælder i kurset, og fungere som rollemodeller ved at gøre brugen af GenAI synlig i undervisningsmaterialer og bedømmelsesforløb, samtidig med at de sikrer overholdelse af databeskyttelsesregler.

Officiel kilde ↗

Sikkerhed og indkøb

Anvendelsesspecifikke begrænsninger

ID CCR Cloud Service Center administrerer institutionelle cloud-AI-værktøjer. Værktøjer, der mangler cloud-vurderinger (herunder tredjepartsværktøjer såsom kommerciel ChatGPT, Claude, DeepSeek, Perplexity, Grok), anvendes på brugerens eget ansvar og må aldrig behandle fortrolige, strengt fortrolige eller beskyttede persondata. Institutionel tilkøbssoftware (såsom Google AI Pro eller M365 Copilot) kræver medarbejderlicensanskaffelse via IT-Shop; specifikke gebyr- og finansieringsordninger afhænger af gældende organisatoriske procedurer.

Officiel kilde ↗

Gør

Aftal den tilladte brug med underviseren eller vejlederen.
Råd fra dette websted
Aftal den tilladte brug med underviseren eller vejlederen.
Omfang
Læs hver citeret regel for dens kursus, faglige enhed, aktivitet og undtagelser. Vejledning på webstedet skaber ikke en forpligtelse for universitetet.
Undervisere afgør, om og hvordan GenAI må anvendes i deres kurser og til respektive bedømmelser. Undervisningsmateriale udarbejdet med GenAI skal underkastes kvalitetskontrol af underviseren.
Oprindelig dokumentation og område
Undervisere afgør, om og hvordan GenAI må anvendes i deres kurser og til respektive bedømmelser. Undervisningsmateriale udarbejdet med GenAI skal underkastes kvalitetskontrol af underviseren.
cc-eth-zurich-0baba14322b53072
Responsibility : As a lecturer you are responsible for the content you provide to your students. This means that teaching materials created with GenAI must be subjected to quality control by you, including checking for possible bias. You determine whether and how GenAI may be used in your courses and for the respective assessments.
Continuous extracted-text characters 3442–3775; extracted text SHA256 087075b276237c15dbd8668988ce5e4cf8168c766f5dbf4fd5de452db7f3dedc; raw source SHA256 05e768573acf7a8f060bb9736d5648b54b290410bc6dcf16621d24d2959c970b; Section: Lecturers — Responsibility; Lines 65–65; characters 3442–3775; excerpt offset 0–333
Officiel universitetskilde ↗
Responsibility: As a lecturer you are responsible for the content you provide to your students. This means that teaching materials created with GenAI must be subjected to quality control by you, including checking for possible bias. You determine whether and how GenAI may be used in your courses and for the respective assessments.
Continuous extracted-text characters 2288–2620; extracted text SHA256 db11c7fd0e4e40100e71eb105e16563f872780b02aa62f442e0cd83752720bb5; raw source SHA256 b883ba3727ad5e0f8af3d12a1846e3a6ecdab325e9ab95b75f80422463cc66ca; Page 1: What does this mean for lecturers? — Responsibility; Lines 41–43; characters 2288–2620; excerpt offset 0–332
Officiel universitetskilde ↗
Tjek den gældende originalitetserklæring og oplysningsaftale.
Råd fra dette websted
Tjek den gældende originalitetserklæring og oplysningsaftale.
Omfang
Læs hver citeret regel for dens kursus, faglige enhed, aktivitet og undtagelser. Vejledning på webstedet skaber ikke en forpligtelse for universitetet.
ETH Zurichs tro- og loveerklæring giver to muligheder: bekræftelse af, at ingen generativ AI blev brugt, eller bekræftelse af, at autoriseret generativ AI blev brugt, og at deklareringen heraf var aftalt med vejlederen. Bachelor- og kandidatspecialer kræver en underskrevet tro- og loveerklæring, mens krav til andre skriftlige bedømmelser afhænger af de gældende akademiske instruktioner. Ph.d.-studerende erklærer uafhængigt forfatterskab gennem tilmeldingsprocessen til ph.d.-eksamenen. Den tilladte brug, deklarationsformatet og eventuelle undtagelser skal følge den relevante vejleders og afdelings krav.
Oprindelig dokumentation og område
ETH Zurichs tro- og loveerklæring giver to muligheder: bekræftelse af, at ingen generativ AI blev brugt, eller bekræftelse af, at autoriseret generativ AI blev brugt, og at deklareringen heraf var aftalt med vejlederen. Bachelor- og kandidatspecialer kræver en underskrevet tro- og loveerklæring, mens krav til andre skriftlige bedømmelser afhænger af de gældende akademiske instruktioner. Ph.d.-studerende erklærer uafhængigt forfatterskab gennem tilmeldingsprocessen til ph.d.-eksamenen. Den tilladte brug, deklarationsformatet og eventuelle undtagelser skal følge den relevante vejleders og afdelings krav.
clm-eth-zurich-originality-declaration-ai-2026
The signed declaration of originality is a component of every written paper or thesis authored during the course of studies. In consultation with the supervisor, one of the following two options must be selected: ☐ I hereby declare that I authored the work in question independently, i.e. that no one helped me to author it. Suggestions from the supervisor regarding language and content are excepted. I used no generative artificial intelligence technologies 1.
Continuous extracted-text characters 27–490; extracted text SHA256 be05265467eaf126b64ee4ddc2a05a35cb319c833a928b5b1cbdcd7c8e78f5b1; raw source SHA256 d3a3ed4457d6c6f5a7dea2d2e084ff201b1f63d8d56af61927e536cb5897f23a; Page 1: Instructions and declaration option 1 (no AI used); Lines 2–8; characters 27–490; excerpt offset 0–463
Officiel universitetskilde ↗
☐ I hereby declare that I authored the work in question independently. In doing so I only used the authorised aids, which included suggestions from the supervisor regarding language and content and generative artificial intelligence technologies. The use of the latter and the respective source declarations proceeded in consultation with the supervisor.
Continuous extracted-text characters 492–847; extracted text SHA256 be05265467eaf126b64ee4ddc2a05a35cb319c833a928b5b1cbdcd7c8e78f5b1; raw source SHA256 d3a3ed4457d6c6f5a7dea2d2e084ff201b1f63d8d56af61927e536cb5897f23a; Page 1: Declaration option 2 (authorised AI used with supervisor agreement); Lines 10–15; characters 492–847; excerpt offset 0–355
Officiel universitetskilde ↗
Concerning the present thesis I hereby declare that: I authored the work in question independently, i.e. that no one helped me to author it. Suggestions from the supervisor regarding language and content are excepted. I used no generative artificial intelligence technologies1 I authored the work in question independently. In doing so I only used the authorised aids, which included suggestions from the supervisor regarding language and content and generative artificial intelligence technologies. The use of the latter and the respective source declarations proceeded in consultation with the supervisor.
Continuous extracted-text characters 1343–1950; extracted text SHA256 e14208dd2e5137b51d71159cd8c473dc19e2c5fcfee0a3b9e9afe820037e3e42; raw source SHA256 2fc181ae19d4630b43464d380b8aa66da7baf4df996ef3b20b3e2ee8db373937; Page 1: Declaration concerning doctoral thesis; Lines 44–49; characters 1343–1950; excerpt offset 0–607
Officiel universitetskilde ↗
At ETH Zurich, students must enclose a signed declaration of originality with their written thesis. This also includes disclosure of the use of artificial intelligence technologies. Bachelor's and Master's students must include a declaration of originality with every written thesis (and, depending on the case, also with every semester paper). Doctoral candidates declare that they have written their doctoral thesis independently as part of their registration for the doctoral examination .
Continuous extracted-text characters 6958–7450; extracted text SHA256 c359d4b3c844b460dd04bd2e8a5022479cfe24f09523b7ca25d74b2e9c1afafb; raw source SHA256 79bad72c13cb4c2710053173b3ed9ce7ca1211954827ca5aa11aed36446c3c02; Section: Transparent declaration — For students and doctoral candidates; Lines 64–66; characters 6958–7450; excerpt offset 0–492
Officiel universitetskilde ↗
Kontrollér AI-genereret information, referencer og citater.
Krav inden for dette område
Kontrollér AI-genereret information, referencer og citater.
Omfang
Læs hver citeret regel for dens kursus, faglige enhed, aktivitet og undtagelser. Vejledning på webstedet skaber ikke en forpligtelse for universitetet.
ETH Librarys vejledning kræver, at forskere forbliver ansvarlige for deres videnskabelige arbejde og verificerer AI-genereret information, referencer og citater. Generative AI-værktøjer kan ikke opføres som videnskabelige forfattere eller anvendes som autoritative videnskabelige kilder. Forskere bør aftale tilladt AI-brug med samarbejdspartnere og følge gældende krav fra institut, finansiering, forlag og tidsskrift. Oplysning om et AI-værktøjs bidrag adskiller sig fra at citere værktøjet som belæg for en videnskabelig påstand.
Oprindelig dokumentation og område
ETH Librarys vejledning kræver, at forskere forbliver ansvarlige for deres videnskabelige arbejde og verificerer AI-genereret information, referencer og citater. Generative AI-værktøjer kan ikke opføres som videnskabelige forfattere eller anvendes som autoritative videnskabelige kilder. Forskere bør aftale tilladt AI-brug med samarbejdspartnere og følge gældende krav fra institut, finansiering, forlag og tidsskrift. Oplysning om et AI-værktøjs bidrag adskiller sig fra at citere værktøjet som belæg for en videnskabelig påstand.
clm-eth-zurich-research-genai-authorship-disclosure
In terms of external page copyright law (Article 2), this does not constitute a new work, which is why these tools cannot be considered as authors or co-authors. Furthermore, as this output is neither reproducible nor reliable, generative AI and AI-based tools are not eligible and citation-worthy sources.
Continuous extracted-text characters 4271–4577; extracted text SHA256 c359d4b3c844b460dd04bd2e8a5022479cfe24f09523b7ca25d74b2e9c1afafb; raw source SHA256 79bad72c13cb4c2710053173b3ed9ce7ca1211954827ca5aa11aed36446c3c02; Section: Authorship — Are AI tools authors?; Lines 55–55; characters 4271–4577; excerpt offset 0–306
Officiel universitetskilde ↗
Taking responsibility means that, as an author, you critically review every output produced by a generative AI or AI-based tool and verify the accuracy of (supposed) facts, sources, references, links, citations, citation styles, etc.
Corrected source binding; Continuous extracted-text characters 5619–5852; extracted text SHA256 c359d4b3c844b460dd04bd2e8a5022479cfe24f09523b7ca25d74b2e9c1afafb; raw source SHA256 79bad72c13cb4c2710053173b3ed9ce7ca1211954827ca5aa11aed36446c3c02; Don'ts Section: Item 1; Lines 60–60; characters 5619–5852; excerpt offset 0–233
Officiel universitetskilde ↗

Undlad

Lad ikke AI erstatte dit eget bedømmelsesarbejde.
Krav inden for dette område
Lad ikke AI erstatte dit eget bedømmelsesarbejde.
Omfang
Læs hver citeret regel for dens kursus, faglige enhed, aktivitet og undtagelser. Vejledning på webstedet skaber ikke en forpligtelse for universitetet.
Studerende er ansvarlige for indholdet af det arbejde, de indleverer. Bedømmelser af præstationer skal gennemføres selvstændigt og personligt; GenAI kan have en supplerende rolle, men må ikke erstatte studerendes egen indsats.
Oprindelig dokumentation og område
Studerende er ansvarlige for indholdet af det arbejde, de indleverer. Bedømmelser af præstationer skal gennemføres selvstændigt og personligt; GenAI kan have en supplerende rolle, men må ikke erstatte studerendes egen indsats.
cc-eth-zurich-b2226b51825107d5
Responsibility : You are responsible for the content of work you submit. Discuss with your lecturer whether and how you may use GenAI and how exactly you should declare it. Be aware to thoroughly check AI generated contributions for correctness and plagiarism . Performance assessments must be conducted independently and personally. GenAI may serve a supplementary role, helping you but not replacing your efforts.
Continuous extracted-text characters 2171–2586; extracted text SHA256 087075b276237c15dbd8668988ce5e4cf8168c766f5dbf4fd5de452db7f3dedc; raw source SHA256 05e768573acf7a8f060bb9736d5648b54b290410bc6dcf16621d24d2959c970b; Section: Students — Responsibility; Lines 61–61; characters 2171–2586; excerpt offset 0–415
Officiel universitetskilde ↗
Responsibility: You are responsible for the content of work you submit. Discuss with your lecturer whether and how you may use GenAI and how exactly you should declare it. Thoroughly check AI generated contributions for plagiarism and correctness. Performance assessments must be conducted independently and personally. GenAI may serve a supplementary role, helping you but not replacing your efforts.
Continuous extracted-text characters 947–1348; extracted text SHA256 db11c7fd0e4e40100e71eb105e16563f872780b02aa62f442e0cd83752720bb5; raw source SHA256 b883ba3727ad5e0f8af3d12a1846e3a6ecdab325e9ab95b75f80422463cc66ca; Page 1: What does this mean for students? — Responsibility; Lines 19–22; characters 947–1348; excerpt offset 0–401
Officiel universitetskilde ↗
Placer ikke strengt fortrolige oplysninger i ekstern cloud-AI.
Krav inden for dette område
Placer ikke strengt fortrolige oplysninger i ekstern cloud-AI.
Omfang
Læs hver citeret regel for dens kursus, faglige enhed, aktivitet og undtagelser. Vejledning på webstedet skaber ikke en forpligtelse for universitetet.
ETH Zurich forbyder behandling af strengt fortrolige oplysninger i eksterne sky-AI-tjenester. Fortrolige oplysninger må kun behandles i tjenester og konfigurationer, der er udtrykkeligt godkendt til den klassifikation. IT-vejledning fra ETH skelner mellem Microsoft 365 Copilot Work-tilstand uden websøgning, som er godkendt til fortrolige data, og Copilot Chat og webtilstand, som er godkendt til interne data. Ikke-vurderede AI-tjenester må ikke bruges til fortrolige ETH-data eller beskyttede personoplysninger.
Oprindelig dokumentation og område
ETH Zurich forbyder behandling af strengt fortrolige oplysninger i eksterne sky-AI-tjenester. Fortrolige oplysninger må kun behandles i tjenester og konfigurationer, der er udtrykkeligt godkendt til den klassifikation. IT-vejledning fra ETH skelner mellem Microsoft 365 Copilot Work-tilstand uden websøgning, som er godkendt til fortrolige data, og Copilot Chat og webtilstand, som er godkendt til interne data. Ikke-vurderede AI-tjenester må ikke bruges til fortrolige ETH-data eller beskyttede personoplysninger.
clm-eth-zurich-ai-data-classification-2026
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Continuous extracted-text characters 503–1028; extracted text SHA256 25f6d4fc047ff5a33ff8b48194071fdb6150dcbb5a64e0513d94abc22c68f958; raw source SHA256 36be2d664bee7d73dc98b2a88fe13b98ea4a029c161a67569491550deb6e402d; Section: Safe use of AI; Lines 9–9; characters 503–1028; excerpt offset 0–525
Officiel universitetskilde ↗
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Continuous extracted-text characters 1224–1687; extracted text SHA256 25f6d4fc047ff5a33ff8b48194071fdb6150dcbb5a64e0513d94abc22c68f958; raw source SHA256 36be2d664bee7d73dc98b2a88fe13b98ea4a029c161a67569491550deb6e402d; Section: Data classification (Strictly confidential & confidential); Lines 10–10; characters 1224–1687; excerpt offset 0–463
Officiel universitetskilde ↗
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Continuous extracted-text characters 1688–1971; extracted text SHA256 25f6d4fc047ff5a33ff8b48194071fdb6150dcbb5a64e0513d94abc22c68f958; raw source SHA256 36be2d664bee7d73dc98b2a88fe13b98ea4a029c161a67569491550deb6e402d; Section: Data classification (Internal data default); Lines 10–10; characters 1688–1971; excerpt offset 0–283
Officiel universitetskilde ↗
Behandl ikke lighedsscreening som et pålideligt bevis for AI-forfatterskab.
Råd fra dette websted
Behandl ikke lighedsscreening som et pålideligt bevis for AI-forfatterskab.
Omfang
Læs hver citeret regel for dens kursus, faglige enhed, aktivitet og undtagelser. Vejledning på webstedet skaber ikke en forpligtelse for universitetet.
ETH-biblioteket stiller Turnitin Originality til rådighed for berettigede professorer og lektorer til tekstlighedskontrol, underlagt dataklassificering og vejledning om ansvarligt selektiv brug. Kun offentlige eller interne dokumenter må uploades. Lighedskontrol er adskilt fra detektion af AI-forfatterskab; ETH Zürich advarer om, at teknisk detektion af AI-genereret output er upålidelig og ikke bør betragtes som afgørende bevis.
Oprindelig dokumentation og område
ETH-biblioteket stiller Turnitin Originality til rådighed for berettigede professorer og lektorer til tekstlighedskontrol, underlagt dataklassificering og vejledning om ansvarligt selektiv brug. Kun offentlige eller interne dokumenter må uploades. Lighedskontrol er adskilt fra detektion af AI-forfatterskab; ETH Zürich advarer om, at teknisk detektion af AI-genereret output er upålidelig og ikke bør betragtes som afgørende bevis.
clm-eth-zurich-turnitin-similarity-and-ai-detection
Please note that access to the plagiarism detection software is exclusively intended for professors and lecturers of ETH Zurich. Students and employees without the lecturer role cannot use Turnitin Originality.
Continuous extracted-text characters 377–587; extracted text SHA256 94d3f7640af7c572e2e1905e5ea8916a25ec1ebc58d9c13b0e1ac0b6093be144; raw source SHA256 fe42a443c6f65d498f8cd583c4947e8ca5354d1cf5a292e6fd721c9ecd20a0e3; Section: Access restrictions (professors & lecturers only); Lines 33–33; characters 377–587; excerpt offset 0–210
Officiel universitetskilde ↗
Note the confidentiality level: Only documents classified as "public" or "internal" may be uploaded to Turnitin Originality for similarity checking.
Continuous extracted-text characters 1577–1725; extracted text SHA256 94d3f7640af7c572e2e1905e5ea8916a25ec1ebc58d9c13b0e1ac0b6093be144; raw source SHA256 fe42a443c6f65d498f8cd583c4947e8ca5354d1cf5a292e6fd721c9ecd20a0e3; Section: Confidentiality level (Public and Internal only); Lines 42–42; characters 1577–1725; excerpt offset 0–148
Officiel universitetskilde ↗
Selective use only in cases of suspicion prevents a culture of mistrust.
Continuous extracted-text characters 2820–2892; extracted text SHA256 94d3f7640af7c572e2e1905e5ea8916a25ec1ebc58d9c13b0e1ac0b6093be144; raw source SHA256 fe42a443c6f65d498f8cd583c4947e8ca5354d1cf5a292e6fd721c9ecd20a0e3; Section: Responsible use (Selective check upon suspicion); Lines 46–46; characters 2820–2892; excerpt offset 0–72
Officiel universitetskilde ↗
False positives: Turnitin Originality may flag sections of texts as plagiarised even if they have been cited correctly. Always check highlighted sections carefully to determine if they have really been plagiarised or if the software has merely produced a false positive.
Continuous extracted-text characters 2437–2707; extracted text SHA256 94d3f7640af7c572e2e1905e5ea8916a25ec1ebc58d9c13b0e1ac0b6093be144; raw source SHA256 fe42a443c6f65d498f8cd583c4947e8ca5354d1cf5a292e6fd721c9ecd20a0e3; Section: False positives (Human review required); Lines 45–45; characters 2437–2707; excerpt offset 0–270
Officiel universitetskilde ↗

Sættes der spørgsmålstegn ved AI-brug?Undersøgelse, mulige udfald og klage

AI-værktøjer og adgang

Microsoft 365 Copilot Chat (Enterprise Data Protection)Institutionel adgang / licensDer gælder betingelser
Log ind med ETH-konto (username@ethz.ch) via M365-abonnement
Microsoft 365 Copilot Chat (Enterprise Data Protection) · Adgangsbetingelser
Alle ETH-medlemmer (studerende og ansatte). Data: Godkendt til INTERNE data (forbudt for fortrolige og strengt fortrolige data). Bedømmelse: Anbefalet til undervisningsformål på IT-siden; brug i kursusarbejde og bedømmelse kræver godkendelse fra underviseren.
Log ind med ETH-konto (username@ethz.ch) via M365-abonnement
Gratis (inkluderet i M365-abonnementet)
Microsoft 365 Copilot Chat (Enterprise Data Protection) · Officiel dokumentation
Kilde 1 ↗
Originalcitat
The service can be accessed in a protected environment via a Microsoft 365 Cloud Subscription , which is provided by ETH Zürich.
Kilde 2 ↗
Originalcitat
The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data .
Kilde 3 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 4 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 5 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 6 ↗
Originalcitat
How Can I Use Microsoft Copilot? × Contact Deutsch Microsoft 365 Manuals and Documentation Guides - Copilot Current: How Can I Use Microsoft Copilot? How Can I Use Microsoft Copilot? This page is only available in English. The Enterprise Plan for Microsoft 365 Copilot has the two following plans: Microsoft 365 Copilot Chat (with Enterprise Protection) available to all ETH users with a Microsoft 365 Subscription Microsoft 365 Copilot (available only for ETH employees) requires a basic license ( Microsoft 365 A5 for faculty ) plus an add-on license that can be acquired through the IT-Shop . Further detailed information can be found here: Microsoft Copilot Manual Navigate to https://copilot.mic
Kilde 7 ↗
Originalcitat
rosoft.com/ Sign in with your ETH Account (username@ ethz.ch ) and credentials. After a successful login, you can use Microsoft Copilot. Getting started with Copilot on Windows - Microsoft Support × Copyright © 2026 Informatikdienste der ETH Zürich • Powered by Scroll Sites and Atlassian Confluence
Google Gemini Chat (Base Offering)Institutionel adgang / licensDer gælder betingelser
ETH Google-konto via Google Workspace Cloud Subscription
Google Gemini Chat (Base Offering) · Adgangsbetingelser
Alle ETH-medlemmer (studerende og ansatte). Data: Ikke vurderet af CISO (OPEN). Værktøjer, der mangler en cloud-vurdering, er ikke frigivet til brug med ETH-data; behandling af fortrolige, strengt fortrolige eller beskyttede personoplysninger er ikke tilladt. Adgang inkluderet under institutionelt Google Workspace-abonnement, men adgang udgør ikke en datasikkerhedsgodkendelse. Bedømmelse: Anbefalet til undervisningsformål på IT-side; brug til kursusarbejde og bedømmelse underlagt individuelle kursus-/underviserregler
ETH Google-konto via Google Workspace Cloud Subscription
Gratis (dækket af institutionelt abonnement)
Google Gemini Chat (Base Offering) · Officiel dokumentation
Kilde 1 ↗
Originalcitat
ETH Zurich offers a Google Workspace Cloud Subscription that allows the service to be used free of charge and in a protected environment. As with Microsoft Copilot, data entered is not used to train the models.
Kilde 2 ↗
Originalcitat
The conversation history retention period is 18 months. In addition, Gemini Chat History can be individually deleted.
Kilde 3 ↗
Originalcitat
Google AI Tools have not yet been assessed by the CISO.
Kilde 4 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 5 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 6 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 7 ↗
Originalcitat
Google Gemini × Contact Deutsch Google Workspace Google Workspace Services Current: Google Gemini Google Gemini This page is only available in English. Gemini apps The Google Workspace subscription at ETH includes the following Gemini apps: Gemini in Workspace apps include in-line features like Help me write in Gmail and the Workspace side panel Gemini Chat is a general-purpose assistant designed to help with a massive range of daily tasks. It can be accessed through gemini.google.com , mobile apps, and Gemini in Chrome. The conversation history retention period is 18 months. In addition, Gemini Chat History can be individually deleted. NotebookLM is Google’s AI-powered research and writing
Kilde 8 ↗
Originalcitat
assistant designed to help synthesize information from given documents. Gemini apps as a core service Gemini Apps as a core service with Google AI Pro for Education Access & license ETH members | Google Workspace subscription ETH employees | Google Workspace subscription + Google AI Pro license Prerequisite Only employees with a Google Workspace subscription can order a Google AI Pro license . App integration with Google Docs, Sheets, etc. Access to features & models with enterprise-grade security & privacy Expanded access to features & models with enterprise-grade security & privacy Cost Included in the Google Workspace subscription Subject to fee Enterprise data protection Cloud assessment
Kilde 9 ↗
Originalcitat
OPEN OPEN Data residency for Gemini in Workspace EUROPE EUROPE Data residency for NotebookLM GLOBAL GLOBAL Check if you have access to Google AI Pro Go to gemini.google.com Check for the Pro badge on the top right. If you find the Pro badge, you have expanded access to features and models. Check if you have access to expanded features & models | Gemini Apps Help Personalization in Gemini apps These features are not available while signed in with ETH accounts (work and school accounts). Get personalization in Gemini Apps | Google Support Guides & Helpful Links Guides & Helpful Links How Can I order Google AI Pro? Google Gemini - Helpful Links × Copyright © 2026 Informatikdienste der ETH Züri
Kilde 10 ↗
Originalcitat
ch • Powered by Scroll Sites and Atlassian Confluence
Google NotebookLM (Workspace Subscription)Institutionel adgang / licensDer gælder betingelser
Log ind med ETH Google-konto via Google Workspace Cloud-abonnement
Google NotebookLM (Workspace Subscription) · Adgangsbetingelser
Alle ETH-medlemmer (studerende og medarbejdere). Data: Ikke vurderet af CISO (OPEN). Værktøjer, som ikke har en cloudvurdering, er ikke frigivet til brug med ETH-data; behandling af fortrolige, strengt fortrolige eller beskyttede personoplysninger er ikke tilladt. Adgang er inkluderet i det institutionelle Google Workspace-abonnement, men adgangen udgør ikke en sikkerhedsgodkendelse af data. Vurdering: Anbefalet til undervisningsformål på IT-siden; brug til opgaver og bedømt arbejde er underlagt underviserens godkendelse.
Log ind med ETH Google-konto via Google Workspace Cloud-abonnement
Gratis (inkluderet i institutionens Google Workspace-abonnement)
Google NotebookLM (Workspace Subscription) · Officiel dokumentation
Kilde 1 ↗
Originalcitat
As with Google Gemini, ETH Zurich offers access via the Google Workspace Cloud subscription . This guarantees the use of the service in a protected environment in which the data entered is not used for training the models.
Kilde 2 ↗
Originalcitat
Google AI Tools have not yet been assessed by the CISO.
Kilde 3 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 4 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 5 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 6 ↗
Originalcitat
AI Tools & Licenses | ETH Zurich Homepage Navigation Search Content Footer Contact Sitemap Header Services Student portal Alumni association Staffnet Contact Login Search Search en EN DE FR (TA) IT (TA) Departments ETH Zurich Select a department Departments D-ARCH: Architecture D-BAUG: Civil, Environmental and Geomatic Engineering D-BSSE: Biosystems Science and Engineering D-INFK: Computer Science D-ITET: Information Technology and Electrical Engineering D-MATL: Department of Materials D-MAVT: Mechanical and Process Engineering D-BIOL: Biology D-CHAB: Chemistry and Applied Biosciences D-MATH: Mathematics D-PHYS: Physics D-EAPS: Earth and Planetary Sciences D-HEST: Health Sciences and Technol
Kilde 7 ↗
Originalcitat
ogy D-USYS: Environmental Systems Science D-MTEC: Management, Technology and Economics D-GESS: Humanities, Social and Political Sciences Language Selection en English Deutsch Français (TA) Italiano (TA) You are here Homepage ETH Zurich Education AI in Teaching and Learning AI Tools & Licenses AI Tools & Licenses This page provides an overview of common generative AI tools, a field that is constantly growing and changing. The use of Microsoft Copilot as well as Google Gemini and NotebookLM is currently recommended, as data-protected access is possible directly via an ETH account. The scope of application of AI goes far beyond the tools listed here. The ETH AI Center is working intensively on
Kilde 8 ↗
Originalcitat
various application scenarios, including in the field of teaching, and is the central hub for artificial intelligence. Some possibilities of application in the field of teaching, which are currently being developed, are listed under projects . You can also find specific information on the direct use of these tools under Assistance for Teaching . Apertus (Image: EPFL, ETH Zurich, CSCS / Molinari Design) The language model external page Apertus was developed in collaboration between ETH Zurich, EPFL, and CSCS (Swiss National Supercomputing Center). It serves as a foundation on which developers and organizations can build future applications, such as chatbots, translation systems, and digital l
Kilde 9 ↗
Originalcitat
earning tools. The entire development process, including architecture, model weights, training data, and methods, is freely accessible and comprehensively documented. Initial impressions of the multilingual model can be gained via the test platform external page Try Apertus . Detailed information about the project can be found on the website of the external page Swiss AI Initiative . Recommended generative AI tools The page on AI-Tools at ETH Zurich provides an up-to-date overview of the growing landscape of various AI tools. The Gemini and NotebookLM offerings from Google, as well as Copilot from Microsoft, which are accessible via an ETH cloud subscription, are available to all ETH employe
Kilde 10 ↗
Originalcitat
es and students via secure access. It is therefore recommended that these tools be used primarily for teaching purposes. Status, October 2025 The external page Microsoft Copilot Website (formerly Bing Chat) is a tool based on generative AI that can generate text and images and is available free of charge. The service can be accessed in a protected environment via a Microsoft 365 Cloud Subscription , which is provided by ETH Zürich. In addition to direct web access, Copilot is also integrated directly into the Edge browser and can also be used via your ETH account. An overview of the entire product range (including M365 Copilot and Github Copilot) is available on the website Microsoft Copilot
Kilde 11 ↗
Originalcitat
from IT Services. The freely accessible version of Microsoft Copilot offers freely available text and image generation, the ‘ protected ’ status (with Microsoft ETH-account ), whereby personal data is not used as training data, a chat history of previously executed queries, the direct integration of an Internet search, links to further resources, uploading a document for specific questioning. The fee-based Pro/Business versions offer prioritised access, GPTs that can be configured for your own needs (‘chat bots’), the integration of Copilot into office and other Microsoft applications. The licence cost for Copilot for M365 is charged on a per-user, per-year basis. All details on the specifi
Kilde 12 ↗
Originalcitat
c and technical information can be found at Microsoft Copilot , the offer overview page of Cloud Services at ETH Zurich. Status, October 2025 The offering external page Google Gemini (formerly Bard) is based on a Large Language Model (LLM) developed by Google itself. It can be used, among other things, to integrate image uploads and voice input to generate texts. ETH Zurich offers a Google Workspace Cloud Subscription that allows the service to be used free of charge and in a protected environment. As with Microsoft Copilot, data entered is not used to train the models. The free version of Google Gemini offers free text generation based on text, image or voice input, a protected status (with
Kilde 13 ↗
Originalcitat
Google ETH account ), whereby personal data is not used by Google as training data, direct integration of an internet search, the chat history of previously executed queries. The advanced version, which is subject to a fee, offers prioritised access and more up-to-date models, direct integration of personal data storage. All the specific details about the ETH Google account in connection with Google Gemini and the associated options can be found at Google AI Tools from ETH Zurich's cloud services. Status, July 2025 The external page NotebookLM application from Google enables users to work and interact with their own resources. It is characterized by a wide range of possible uses. In additio
Kilde 14 ↗
Originalcitat
n, a customized podcast can be created that provides alternative access to the content provided. Notebooks can be shared within ETH Zurich. This enables access to resources within groups or for students as part of learning units. As with Google Gemini, ETH Zurich offers access via the Google Workspace Cloud subscription . This guarantees the use of the service in a protected environment in which the data entered is not used for training the models. The free version of NotebookLM offers the ability to upload your own documents and link to websites, create summaries and interact with content, and generate podcasts. The paid Pro version offers expanded resource integration, access to newer mode
Kilde 15 ↗
Originalcitat
ls, and advanced customisation options. Further information on the offerings can be found at Google AI Tools in the ETH Zurich IT knowledge database. Additional AI tools The following non-exhaustive list shows a selection of these tools that are currently well known or have proven to be particularly useful. Status, December 2024 OpenAI was the first player on the market with external page ChatGPT and was able to establish itself across the board very quickly. The further development of the Large Language Models (LLM), the refinement of the options for interacting with the model and the range of products on offer are constantly expanding. With the introduction of GPTs (custom GPTs, personal ‘
Kilde 16 ↗
Originalcitat
chat bots’) in autumn 2023, it also became possible to enhance the outputs of generative AI with your own data and configure them to suit your own needs. This offer is subject to certain restrictions and in some cases is only possible via the paid Plus version. The freely accessible version of ChaptGPT offers freely available text generation, a chat history of previously executed queries ( note : data is stored on the OpenAI server). The Plus/Team versions, which are subject to a charge, offer prioritised access and more up-to-date models, image generation, GPTs (‘chat bots’) that can be configured for your own needs. A personal account must always be created to access ChatGPT. It is strongl
Kilde 17 ↗
Originalcitat
y recommended to switch off the option "Improve the model for everyone" so that inserted texts and data are not used for training the model (see external page Data Controls FAQ von OpenAI). Status, December 2024 Anthropic, the company behind the product external page Claude.AI , has made a name for itself in recent months with its highly advanced Large Language Model (LLM). This product also allows users to upload images and integrate multiple documents (upload limit in the free version). The freely accessible version of Claude.AI offers freely available text generation based on text, upload of your own documents, a chat history of previously executed queries ( note : data is stored on the s
Kilde 18 ↗
Originalcitat
erver). The chargeable Pro-/Team-version offers prioritised access and more up-to-date models, organisation of projects as well as document mangament. In order to access Claude.AI, a personal account must be created in any case. Anthropic guarantees that the data used for the interaction with the LLM will not be used to train their models (see Anthropic's external page Data Handling & Retention ). Below is a list of some AI tools that are currently also frequently used and supplement the tools above. The list does not claim to be exhaustive. external page Perplexity AI focusses on correct references to further resources. external page DeepSeek and external page Grok are other well known and
Kilde 19 ↗
Originalcitat
often discussed GenAI tools. external page There's an AI for That offers a search engine for every conceivable AI tool. A number of AI-based tools have already become established, particularly in scientific work and literature research. A brief overview can be found here. Connected Papers is a literature research tool (premium version via the ETH Zurich account). Scopus AI and Scite , AI-powered research tools available at ETH Zurich. external page Open Knowledge Maps , an AI-based search engine for science. external page ResearchRabbit provides support in organising and searching for literature. Information security and artificial intelligence In order to make targeted but also responsible
Kilde 20 ↗
Originalcitat
use of the possibilities offered by generative AI technologies while minimizing risks such as data protection violations and ethical conflicts, IT Services Information Security has published guidelines for the safe use of AI at ETH Zurich . The protection of sensitive and personal data is a top priority. In addition, it is important to always keep an eye on the critical review of content and compliance with legal frameworks. Footer Recommended links Media information Search Keyword or person Follow us Services Student portal Alumni association Staffnet Contact Login Departments D-ARCH Architecture D-BAUG Civil, Environmental and Geomatic Engineering D-BIOL Biology D-BSSE Biosystems Science a
Kilde 21 ↗
Originalcitat
nd Engineering D-CHAB Chemistry and Applied Biosciences D-EAPS Earth and Planetary Sciences D-GESS Humanities, Social and Political Sciences D-HEST Health Sciences and Technology D-INFK Computer Science D-ITET Information Technology and Electrical Engineering D-MATH Mathematics D-MATL Department of Materials D-MAVT Mechanical and Process Engineering D-MTEC Management, Technology and Economics D-PHYS Physics D-USYS Environmental Systems Science Table of contents and legal Sitemap Imprint Accessibility Statement Disclaimer & Copyright Data protection © 2026 Eidgenössische Technische Hochschule Zürich JavaScript has been disabled in your browser

These tools have audience, mode or purpose limits; access is not established for all students.

Tools for specific users and service modes

Microsoft 365 Copilot (Work Mode - Enterprise)Institutionel adgang / licensDer gælder betingelser
ETH-medarbejderkonto med M365-abonnement og tilføjelseslicens erhvervet via IT-Shop (M365 A5-licens for undervisere, nævnt i IT-vejledninger; specifikke detaljer om licenstildeling er ukendte)
Microsoft 365 Copilot (Work Mode - Enterprise) · Adgangsbetingelser
Kun ETH-ansatte. Data: Godkendt til CONFIDENTIAL data, når websøgning er deaktiveret (forbudt for strengt fortrolige data) Bedømmelse: ukendt (IT-siden definerer adgang for ansatte og regler for fortrolige data; specificerer ikke tilladelser til undervisning/bedømmelse; underlagt gældende kursusregler)
ETH-medarbejderkonto med M365-abonnement og tilføjelseslicens erhvervet via IT-Shop (M365 A5-licens for undervisere, nævnt i IT-vejledninger; specifikke detaljer om licenstildeling er ukendte)
Pålægges gebyr (den officielle side anfører tillægslicens via IT-Shop; specifikt gebyrbeløb og afregningsperiode er ukendte)
Microsoft 365 Copilot (Work Mode - Enterprise) · Officiel dokumentation
Kilde 1 ↗
Originalcitat
The service can be accessed in a protected environment via a Microsoft 365 Cloud Subscription , which is provided by ETH Zürich.
Kilde 2 ↗
Originalcitat
The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data .
Kilde 3 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 4 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 5 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 6 ↗
Originalcitat
How Can I Use Microsoft Copilot? × Contact Deutsch Microsoft 365 Manuals and Documentation Guides - Copilot Current: How Can I Use Microsoft Copilot? How Can I Use Microsoft Copilot? This page is only available in English. The Enterprise Plan for Microsoft 365 Copilot has the two following plans: Microsoft 365 Copilot Chat (with Enterprise Protection) available to all ETH users with a Microsoft 365 Subscription Microsoft 365 Copilot (available only for ETH employees) requires a basic license ( Microsoft 365 A5 for faculty ) plus an add-on license that can be acquired through the IT-Shop . Further detailed information can be found here: Microsoft Copilot Manual Navigate to https://copilot.mic
Kilde 7 ↗
Originalcitat
rosoft.com/ Sign in with your ETH Account (username@ ethz.ch ) and credentials. After a successful login, you can use Microsoft Copilot. Getting started with Copilot on Windows - Microsoft Support × Copyright © 2026 Informatikdienste der ETH Zürich • Powered by Scroll Sites and Atlassian Confluence
Microsoft 365 Copilot (Web Mode)Institutionel adgang / licensDer gælder betingelser
M365 Copilot-licens med websøgningskontakten aktiveret
Microsoft 365 Copilot (Web Mode) · Adgangsbetingelser
ETH-ansatte med licens til M365 Copilot. Data: Godkendt kun til INTERNE data (webssøgninger forlader tenant-grænsen). Bedømmelse: ukendt (underlagt gældende kursusregler)
M365 Copilot-licens med websøgningskontakten aktiveret
Inkluderet i medarbejderlicens til M365 Copilot (specifikt tilkøbsgebyr ukendt)
Microsoft 365 Copilot (Web Mode) · Officiel dokumentation
Kilde 1 ↗
Originalcitat
The service can be accessed in a protected environment via a Microsoft 365 Cloud Subscription , which is provided by ETH Zürich.
Kilde 2 ↗
Originalcitat
The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data .
Kilde 3 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 4 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 5 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 6 ↗
Originalcitat
How Can I Use Microsoft Copilot? × Contact Deutsch Microsoft 365 Manuals and Documentation Guides - Copilot Current: How Can I Use Microsoft Copilot? How Can I Use Microsoft Copilot? This page is only available in English. The Enterprise Plan for Microsoft 365 Copilot has the two following plans: Microsoft 365 Copilot Chat (with Enterprise Protection) available to all ETH users with a Microsoft 365 Subscription Microsoft 365 Copilot (available only for ETH employees) requires a basic license ( Microsoft 365 A5 for faculty ) plus an add-on license that can be acquired through the IT-Shop . Further detailed information can be found here: Microsoft Copilot Manual Navigate to https://copilot.mic
Kilde 7 ↗
Originalcitat
rosoft.com/ Sign in with your ETH Account (username@ ethz.ch ) and credentials. After a successful login, you can use Microsoft Copilot. Getting started with Copilot on Windows - Microsoft Support × Copyright © 2026 Informatikdienste der ETH Zürich • Powered by Scroll Sites and Atlassian Confluence
Google AI Pro for Education (Add-on)Institutionel adgang / licensDer gælder betingelser
ETH Google-konto med Google AI Pro-licens (bestilt via IT-Shop; ansatte med Google Workspace-abonnement; specifikke godkendelses-/faktureringskrav ukendte)
Google AI Pro for Education (Add-on) · Adgangsbetingelser
Kun ETH-ansatte (videnskabeligt og administrativt personale med Google Workspace-abonnement). Data: Ikke vurderet af CISO (OPEN). Ikke frigivet til brug med ETH-data; behandling af fortrolige, strengt fortrolige eller beskyttede persondata er ikke tilladt. Vurdering: ukendt (it-siden specificerer kun medarbejderadgang og licensering; definerer ikke undervisnings- eller vurderingstilladelse; underlagt gældende regler fra vejleder og kursus)
ETH Google-konto med Google AI Pro-licens (bestilt via IT-Shop; ansatte med Google Workspace-abonnement; specifikke godkendelses-/faktureringskrav ukendte)
Gebyrpligtigt („Gebyrpligtigt“ fremgår af den officielle dokumentation; specifik faktureringsperiode og fordeling på omkostningssted er ukendt)
Google AI Pro for Education (Add-on) · Officiel dokumentation
Kilde 1 ↗
Originalcitat
ETH Zurich offers a Google Workspace Cloud Subscription that allows the service to be used free of charge and in a protected environment. As with Microsoft Copilot, data entered is not used to train the models.
Kilde 2 ↗
Originalcitat
The conversation history retention period is 18 months. In addition, Gemini Chat History can be individually deleted.
Kilde 3 ↗
Originalcitat
Google AI Tools have not yet been assessed by the CISO.
Kilde 4 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 5 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 6 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 7 ↗
Originalcitat
Google Gemini × Contact Deutsch Google Workspace Google Workspace Services Current: Google Gemini Google Gemini This page is only available in English. Gemini apps The Google Workspace subscription at ETH includes the following Gemini apps: Gemini in Workspace apps include in-line features like Help me write in Gmail and the Workspace side panel Gemini Chat is a general-purpose assistant designed to help with a massive range of daily tasks. It can be accessed through gemini.google.com , mobile apps, and Gemini in Chrome. The conversation history retention period is 18 months. In addition, Gemini Chat History can be individually deleted. NotebookLM is Google’s AI-powered research and writing
Kilde 8 ↗
Originalcitat
assistant designed to help synthesize information from given documents. Gemini apps as a core service Gemini Apps as a core service with Google AI Pro for Education Access & license ETH members | Google Workspace subscription ETH employees | Google Workspace subscription + Google AI Pro license Prerequisite Only employees with a Google Workspace subscription can order a Google AI Pro license . App integration with Google Docs, Sheets, etc. Access to features & models with enterprise-grade security & privacy Expanded access to features & models with enterprise-grade security & privacy Cost Included in the Google Workspace subscription Subject to fee Enterprise data protection Cloud assessment
Kilde 9 ↗
Originalcitat
OPEN OPEN Data residency for Gemini in Workspace EUROPE EUROPE Data residency for NotebookLM GLOBAL GLOBAL Check if you have access to Google AI Pro Go to gemini.google.com Check for the Pro badge on the top right. If you find the Pro badge, you have expanded access to features and models. Check if you have access to expanded features & models | Gemini Apps Help Personalization in Gemini apps These features are not available while signed in with ETH accounts (work and school accounts). Get personalization in Gemini Apps | Google Support Guides & Helpful Links Guides & Helpful Links How Can I order Google AI Pro? Google Gemini - Helpful Links × Copyright © 2026 Informatikdienste der ETH Züri
Kilde 10 ↗
Originalcitat
ch • Powered by Scroll Sites and Atlassian Confluence
Google AI Studio PilotInstitutionel adgang / licensDer gælder betingelser
Anmod om deltagelse i pilotprojektet via IT Service Desk (servicedesk@id.ethz.ch)
Google AI Studio Pilot · Adgangsbetingelser
ETH-medlemmer med Google Workspace-abonnement (begrænset pilotprojekt). Data: Kun pilotadgang; ikke vurderet af CISO (OPEN); ikke frigivet til ETH-fortrolige, strengt fortrolige eller beskyttede personoplysninger. Vurdering: ukendt (IT-siden angiver kun konto- og pilotberettigelse; angiver ikke godkendelse eller forbud vedrørende kursusarbejde eller eksamen; underlagt gældende kursusregler)
Anmod om deltagelse i pilotprojektet via IT Service Desk (servicedesk@id.ethz.ch)
ukendt
Google AI Studio Pilot · Officiel dokumentation
Kilde 1 ↗
Originalcitat
ETH Zurich offers a Google Workspace Cloud Subscription that allows the service to be used free of charge and in a protected environment. As with Microsoft Copilot, data entered is not used to train the models.
Kilde 2 ↗
Originalcitat
The conversation history retention period is 18 months. In addition, Gemini Chat History can be individually deleted.
Kilde 3 ↗
Originalcitat
Google AI Tools have not yet been assessed by the CISO.
Kilde 4 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 5 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 6 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 7 ↗
Originalcitat
Google Gemini × Contact Deutsch Google Workspace Google Workspace Services Current: Google Gemini Google Gemini This page is only available in English. Gemini apps The Google Workspace subscription at ETH includes the following Gemini apps: Gemini in Workspace apps include in-line features like Help me write in Gmail and the Workspace side panel Gemini Chat is a general-purpose assistant designed to help with a massive range of daily tasks. It can be accessed through gemini.google.com , mobile apps, and Gemini in Chrome. The conversation history retention period is 18 months. In addition, Gemini Chat History can be individually deleted. NotebookLM is Google’s AI-powered research and writing
Kilde 8 ↗
Originalcitat
assistant designed to help synthesize information from given documents. Gemini apps as a core service Gemini Apps as a core service with Google AI Pro for Education Access & license ETH members | Google Workspace subscription ETH employees | Google Workspace subscription + Google AI Pro license Prerequisite Only employees with a Google Workspace subscription can order a Google AI Pro license . App integration with Google Docs, Sheets, etc. Access to features & models with enterprise-grade security & privacy Expanded access to features & models with enterprise-grade security & privacy Cost Included in the Google Workspace subscription Subject to fee Enterprise data protection Cloud assessment
Kilde 9 ↗
Originalcitat
OPEN OPEN Data residency for Gemini in Workspace EUROPE EUROPE Data residency for NotebookLM GLOBAL GLOBAL Check if you have access to Google AI Pro Go to gemini.google.com Check for the Pro badge on the top right. If you find the Pro badge, you have expanded access to features and models. Check if you have access to expanded features & models | Gemini Apps Help Personalization in Gemini apps These features are not available while signed in with ETH accounts (work and school accounts). Get personalization in Gemini Apps | Google Support Guides & Helpful Links Guides & Helpful Links How Can I order Google AI Pro? Google Gemini - Helpful Links × Copyright © 2026 Informatikdienste der ETH Züri
Kilde 10 ↗
Originalcitat
ch • Powered by Scroll Sites and Atlassian Confluence
Google Vertex AI (Cloud Project)Institutionel adgang / licensDer gælder betingelser
Google Cloud Project-abonnement administreret via ID CCR Cloud Service Center
Google Vertex AI (Cloud Project) · Adgangsbetingelser
ETH-medarbejdere med Google Workspace-abonnement og Google Cloud Project. Data: Kræver en projektspecifik skyvurdering, før projektdata behandles; ikke-vurderede skytjenester må ikke behandle fortrolige eller strengt fortrolige ETH-data. Vurdering: ukendt (IT-siden angiver kun medarbejderadgang og Google Project-abonnement; den definerer ikke eksklusiv brug eller tilladelser til undervisning og bedømmelse; underlagt gældende kursusregler)
Google Cloud Project-abonnement administreret via ID CCR Cloud Service Center
ukendt
Google Vertex AI (Cloud Project) · Officiel dokumentation
Kilde 1 ↗
Originalcitat
ETH Zurich offers a Google Workspace Cloud Subscription that allows the service to be used free of charge and in a protected environment. As with Microsoft Copilot, data entered is not used to train the models.
Kilde 2 ↗
Originalcitat
The conversation history retention period is 18 months. In addition, Gemini Chat History can be individually deleted.
Kilde 3 ↗
Originalcitat
Google AI Tools have not yet been assessed by the CISO.
Kilde 4 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 5 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 6 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 7 ↗
Originalcitat
Google Gemini × Contact Deutsch Google Workspace Google Workspace Services Current: Google Gemini Google Gemini This page is only available in English. Gemini apps The Google Workspace subscription at ETH includes the following Gemini apps: Gemini in Workspace apps include in-line features like Help me write in Gmail and the Workspace side panel Gemini Chat is a general-purpose assistant designed to help with a massive range of daily tasks. It can be accessed through gemini.google.com , mobile apps, and Gemini in Chrome. The conversation history retention period is 18 months. In addition, Gemini Chat History can be individually deleted. NotebookLM is Google’s AI-powered research and writing
Kilde 8 ↗
Originalcitat
assistant designed to help synthesize information from given documents. Gemini apps as a core service Gemini Apps as a core service with Google AI Pro for Education Access & license ETH members | Google Workspace subscription ETH employees | Google Workspace subscription + Google AI Pro license Prerequisite Only employees with a Google Workspace subscription can order a Google AI Pro license . App integration with Google Docs, Sheets, etc. Access to features & models with enterprise-grade security & privacy Expanded access to features & models with enterprise-grade security & privacy Cost Included in the Google Workspace subscription Subject to fee Enterprise data protection Cloud assessment
Kilde 9 ↗
Originalcitat
OPEN OPEN Data residency for Gemini in Workspace EUROPE EUROPE Data residency for NotebookLM GLOBAL GLOBAL Check if you have access to Google AI Pro Go to gemini.google.com Check for the Pro badge on the top right. If you find the Pro badge, you have expanded access to features and models. Check if you have access to expanded features & models | Gemini Apps Help Personalization in Gemini apps These features are not available while signed in with ETH accounts (work and school accounts). Get personalization in Gemini Apps | Google Support Guides & Helpful Links Guides & Helpful Links How Can I order Google AI Pro? Google Gemini - Helpful Links × Copyright © 2026 Informatikdienste der ETH Züri
Kilde 10 ↗
Originalcitat
ch • Powered by Scroll Sites and Atlassian Confluence
Turnitin OriginalityOfficiel anbefalingDer gælder betingelser
Adgang via ETH-konto administreret af ETH Library (turnitin.ethz.ch)
Turnitin Originality · Adgangsbetingelser
Kun professorer og lektorer (studerende og ikke-undervisere har ingen direkte adgang). Data: Kun offentlige og interne dokumenter (fortrolige eller strengt fortrolige dokumenter forbudt) Bedømmelse: Selektiv screening ved mistanke anbefales; menneskelig gennemgang påkrævet ved falske positive; bredt forbud fra direktionen fremgår ikke af den officielle side
Adgang via ETH-konto administreret af ETH Library (turnitin.ethz.ch)
ukendt
Turnitin Originality · Officiel dokumentation
Kilde 1 ↗
Originalcitat
Please note that access to the plagiarism detection software is exclusively intended for professors and lecturers of ETH Zurich. Students and employees without the lecturer role cannot use Turnitin Originality.
Kilde 2 ↗
Originalcitat
Note the confidentiality level: Only documents classified as "public" or "internal" may be uploaded to Turnitin Originality for similarity checking.
Kilde 3 ↗
Originalcitat
Selective use only in cases of suspicion prevents a culture of mistrust.
Kilde 4 ↗
Originalcitat
False positives: Turnitin Originality may flag sections of texts as plagiarised even if they have been cited correctly. Always check highlighted sections carefully to determine if they have really been plagiarised or if the software has merely produced a false positive.
Kilde 5 ↗
Originalcitat
Plagiarism check – ETH Library | ETH Zurich Homepage Navigation Search Content Footer Contact Sitemap ETH Library Services Student portal Alumni association Staffnet Login Search Search en EN DE Departments ETH Zurich ETH Library Language Selection en English Deutsch You are here Homepage Scientific writing Plagiarism and academic integrity Plagiarism check Plagiarism check Please note that access to the plagiarism detection software is exclusively intended for professors and lecturers of ETH Zurich. Students and employees without the lecturer role cannot use Turnitin Originality. (Access to Turnitin Originality (with ETH account)) Turnitin Originality is a web-based text similarity detectio
Kilde 6 ↗
Originalcitat
n software. Submitted documents are compared against an extensive database of academic papers (both published and unpublished), books, journals, and other sources: Some features: Doc-to-doc comparison: direct comparison of a document with one or up to 5 other documents. Detecting plagiarism: Turnitin Originality detects similarities in texts including paraphrasing, citations and references. Originality reports: The tool generates an originality report that highlights all the areas of the document that match the sources in the database. The report also shows the percentage of similarity and the matching sources. Checking quotations and references: Turnitin Originality can check citations and
Kilde 7 ↗
Originalcitat
references in submitted documents to ensure that they are properly formatted and incorporated correctly. There are a few things to keep in mind when using Turnitin Originality: Note the confidentiality level: Only documents classified as "public" or "internal" may be uploaded to Turnitin Originality for similarity checking. For more details, see the Download terms and conditions of use (PDF, 143 KB) . Understand the limitations: Turnitin Originality is not infallible. It relies on its source database and may not detect all instances of plagiarism, for example, if the source is not in the database. Turnitin Originality is not a substitute for teaching proper citation and referencing practices
Kilde 8 ↗
Originalcitat
. Instructors should guide students on correct citation and referencing. Here, the ETH Library offers support. Watch out for false positives: Turnitin Originality can mark text as plagiarised even if it is properly cited. It is important to carefully review the flagged text to determine if it is indeed plagiarised or a false positive. False positives: Turnitin Originality may flag sections of texts as plagiarised even if they have been cited correctly. Always check highlighted sections carefully to determine if they have really been plagiarised or if the software has merely produced a false positive. Use the software responsibly: Plagiarism detection software should be used in a responsible
Kilde 9 ↗
Originalcitat
and ethical manner. Selective use only in cases of suspicion prevents a culture of mistrust. The following use cases are possible: Plagiarism check of written work of students from the teaching context such as term papers and bachelor's/master's theses: The plagiarism check is carried out by supervisors (usually professors or lecturers) if plagiarism is suspected. If the suspicion is confirmed, it is communicated by the supervisors to the ETH's Office of Legal Affairs in Teaching and further processed together with them. Plagiarism check of written work that is produced in the context of a doctorate: Term papers, papers, abstracts, doctoral theses. The examination is carried out by superviso
Kilde 10 ↗
Originalcitat
rs (professors or lecturers) if plagiarism is suspected. If the suspicion is confirmed, it will be communicated by the supervisors to the ETH Research Ethics via trusted persons and followed up. Plagiarism check of texts produced in the context of research: Papers, abstracts, habilitation theses, etc. The check is carried out by members of the department (mostly professors or lecturers) if plagiarism is suspected. If the suspicion is confirmed: proceed as described in the previous point. Related links Avoid plagiarism by knowing how to cite correctly Download Checklist correct citing (PDF, 63 KB) Preventing plagiarism – citing correctly: also in times of AI Prevention of plagiarism Scientifi
Kilde 11 ↗
Originalcitat
c integrity Contact Plagiarism check Miklos Frank & Raymond Grenacher Footer Recommended links Conditions of Use, Fees Frequently asked questions Search Keyword or person Follow us Services Student portal Alumni association Staffnet Login Contact Phone +41 44 632 21 35 ETH Library Rämistrasse 101 8092 Zurich Show more Show less Departments D-ARCH Architecture D-BAUG Civil, Environmental and Geomatic Engineering D-BIOL Biology D-BSSE Biosystems Science and Engineering D-CHAB Chemistry and Applied Biosciences D-EAPS Earth and Planetary Sciences D-GESS Humanities, Social and Political Sciences D-HEST Health Sciences and Technology D-INFK Computer Science D-ITET Information Technology and Electr
Kilde 12 ↗
Originalcitat
ical Engineering D-MATH Mathematics D-MATL Department of Materials D-MAVT Mechanical and Process Engineering D-MTEC Management, Technology and Economics D-PHYS Physics D-USYS Environmental Systems Science Table of contents and legal Sitemap Imprint Accessibility Statement Disclaimer & Copyright Data protection © 2026 Eidgenössische Technische Hochschule Zürich Chat x JavaScript has been disabled in your browser
ApertusOfficiel anbefalingDer gælder betingelser
Åben webtestplatform (Try Apertus) og frit downloadbare vægte og data
Apertus · Adgangsbetingelser
Forskere, udviklere, studerende og organisationer (åben model / Try Apertus-platform). Data: ukendt. Vurdering: ukendt.
Åben webtestplatform (Try Apertus) og frit downloadbare vægte og data
Gratis (open source-modelvægte og webplatform)
Apertus · Officiel dokumentation
Kilde 1 ↗
Originalcitat
AI Tools & Licenses | ETH Zurich Homepage Navigation Search Content Footer Contact Sitemap Header Services Student portal Alumni association Staffnet Contact Login Search Search en EN DE FR (TA) IT (TA) Departments ETH Zurich Select a department Departments D-ARCH: Architecture D-BAUG: Civil, Environmental and Geomatic Engineering D-BSSE: Biosystems Science and Engineering D-INFK: Computer Science D-ITET: Information Technology and Electrical Engineering D-MATL: Department of Materials D-MAVT: Mechanical and Process Engineering D-BIOL: Biology D-CHAB: Chemistry and Applied Biosciences D-MATH: Mathematics D-PHYS: Physics D-EAPS: Earth and Planetary Sciences D-HEST: Health Sciences and Technol
Kilde 2 ↗
Originalcitat
ogy D-USYS: Environmental Systems Science D-MTEC: Management, Technology and Economics D-GESS: Humanities, Social and Political Sciences Language Selection en English Deutsch Français (TA) Italiano (TA) You are here Homepage ETH Zurich Education AI in Teaching and Learning AI Tools & Licenses AI Tools & Licenses This page provides an overview of common generative AI tools, a field that is constantly growing and changing. The use of Microsoft Copilot as well as Google Gemini and NotebookLM is currently recommended, as data-protected access is possible directly via an ETH account. The scope of application of AI goes far beyond the tools listed here. The ETH AI Center is working intensively on
Kilde 3 ↗
Originalcitat
various application scenarios, including in the field of teaching, and is the central hub for artificial intelligence. Some possibilities of application in the field of teaching, which are currently being developed, are listed under projects . You can also find specific information on the direct use of these tools under Assistance for Teaching . Apertus (Image: EPFL, ETH Zurich, CSCS / Molinari Design) The language model external page Apertus was developed in collaboration between ETH Zurich, EPFL, and CSCS (Swiss National Supercomputing Center). It serves as a foundation on which developers and organizations can build future applications, such as chatbots, translation systems, and digital l
Kilde 4 ↗
Originalcitat
earning tools. The entire development process, including architecture, model weights, training data, and methods, is freely accessible and comprehensively documented. Initial impressions of the multilingual model can be gained via the test platform external page Try Apertus . Detailed information about the project can be found on the website of the external page Swiss AI Initiative . Recommended generative AI tools The page on AI-Tools at ETH Zurich provides an up-to-date overview of the growing landscape of various AI tools. The Gemini and NotebookLM offerings from Google, as well as Copilot from Microsoft, which are accessible via an ETH cloud subscription, are available to all ETH employe
Kilde 5 ↗
Originalcitat
es and students via secure access. It is therefore recommended that these tools be used primarily for teaching purposes. Status, October 2025 The external page Microsoft Copilot Website (formerly Bing Chat) is a tool based on generative AI that can generate text and images and is available free of charge. The service can be accessed in a protected environment via a Microsoft 365 Cloud Subscription , which is provided by ETH Zürich. In addition to direct web access, Copilot is also integrated directly into the Edge browser and can also be used via your ETH account. An overview of the entire product range (including M365 Copilot and Github Copilot) is available on the website Microsoft Copilot
Kilde 6 ↗
Originalcitat
from IT Services. The freely accessible version of Microsoft Copilot offers freely available text and image generation, the ‘ protected ’ status (with Microsoft ETH-account ), whereby personal data is not used as training data, a chat history of previously executed queries, the direct integration of an Internet search, links to further resources, uploading a document for specific questioning. The fee-based Pro/Business versions offer prioritised access, GPTs that can be configured for your own needs (‘chat bots’), the integration of Copilot into office and other Microsoft applications. The licence cost for Copilot for M365 is charged on a per-user, per-year basis. All details on the specifi
Kilde 7 ↗
Originalcitat
c and technical information can be found at Microsoft Copilot , the offer overview page of Cloud Services at ETH Zurich. Status, October 2025 The offering external page Google Gemini (formerly Bard) is based on a Large Language Model (LLM) developed by Google itself. It can be used, among other things, to integrate image uploads and voice input to generate texts. ETH Zurich offers a Google Workspace Cloud Subscription that allows the service to be used free of charge and in a protected environment. As with Microsoft Copilot, data entered is not used to train the models. The free version of Google Gemini offers free text generation based on text, image or voice input, a protected status (with
Kilde 8 ↗
Originalcitat
Google ETH account ), whereby personal data is not used by Google as training data, direct integration of an internet search, the chat history of previously executed queries. The advanced version, which is subject to a fee, offers prioritised access and more up-to-date models, direct integration of personal data storage. All the specific details about the ETH Google account in connection with Google Gemini and the associated options can be found at Google AI Tools from ETH Zurich's cloud services. Status, July 2025 The external page NotebookLM application from Google enables users to work and interact with their own resources. It is characterized by a wide range of possible uses. In additio
Kilde 9 ↗
Originalcitat
n, a customized podcast can be created that provides alternative access to the content provided. Notebooks can be shared within ETH Zurich. This enables access to resources within groups or for students as part of learning units. As with Google Gemini, ETH Zurich offers access via the Google Workspace Cloud subscription . This guarantees the use of the service in a protected environment in which the data entered is not used for training the models. The free version of NotebookLM offers the ability to upload your own documents and link to websites, create summaries and interact with content, and generate podcasts. The paid Pro version offers expanded resource integration, access to newer mode
Kilde 10 ↗
Originalcitat
ls, and advanced customisation options. Further information on the offerings can be found at Google AI Tools in the ETH Zurich IT knowledge database. Additional AI tools The following non-exhaustive list shows a selection of these tools that are currently well known or have proven to be particularly useful. Status, December 2024 OpenAI was the first player on the market with external page ChatGPT and was able to establish itself across the board very quickly. The further development of the Large Language Models (LLM), the refinement of the options for interacting with the model and the range of products on offer are constantly expanding. With the introduction of GPTs (custom GPTs, personal ‘
Kilde 11 ↗
Originalcitat
chat bots’) in autumn 2023, it also became possible to enhance the outputs of generative AI with your own data and configure them to suit your own needs. This offer is subject to certain restrictions and in some cases is only possible via the paid Plus version. The freely accessible version of ChaptGPT offers freely available text generation, a chat history of previously executed queries ( note : data is stored on the OpenAI server). The Plus/Team versions, which are subject to a charge, offer prioritised access and more up-to-date models, image generation, GPTs (‘chat bots’) that can be configured for your own needs. A personal account must always be created to access ChatGPT. It is strongl
Kilde 12 ↗
Originalcitat
y recommended to switch off the option "Improve the model for everyone" so that inserted texts and data are not used for training the model (see external page Data Controls FAQ von OpenAI). Status, December 2024 Anthropic, the company behind the product external page Claude.AI , has made a name for itself in recent months with its highly advanced Large Language Model (LLM). This product also allows users to upload images and integrate multiple documents (upload limit in the free version). The freely accessible version of Claude.AI offers freely available text generation based on text, upload of your own documents, a chat history of previously executed queries ( note : data is stored on the s
Kilde 13 ↗
Originalcitat
erver). The chargeable Pro-/Team-version offers prioritised access and more up-to-date models, organisation of projects as well as document mangament. In order to access Claude.AI, a personal account must be created in any case. Anthropic guarantees that the data used for the interaction with the LLM will not be used to train their models (see Anthropic's external page Data Handling & Retention ). Below is a list of some AI tools that are currently also frequently used and supplement the tools above. The list does not claim to be exhaustive. external page Perplexity AI focusses on correct references to further resources. external page DeepSeek and external page Grok are other well known and
Kilde 14 ↗
Originalcitat
often discussed GenAI tools. external page There's an AI for That offers a search engine for every conceivable AI tool. A number of AI-based tools have already become established, particularly in scientific work and literature research. A brief overview can be found here. Connected Papers is a literature research tool (premium version via the ETH Zurich account). Scopus AI and Scite , AI-powered research tools available at ETH Zurich. external page Open Knowledge Maps , an AI-based search engine for science. external page ResearchRabbit provides support in organising and searching for literature. Information security and artificial intelligence In order to make targeted but also responsible
Kilde 15 ↗
Originalcitat
use of the possibilities offered by generative AI technologies while minimizing risks such as data protection violations and ethical conflicts, IT Services Information Security has published guidelines for the safe use of AI at ETH Zurich . The protection of sensitive and personal data is a top priority. In addition, it is important to always keep an eye on the critical review of content and compliance with legal frameworks. Footer Recommended links Media information Search Keyword or person Follow us Services Student portal Alumni association Staffnet Contact Login Departments D-ARCH Architecture D-BAUG Civil, Environmental and Geomatic Engineering D-BIOL Biology D-BSSE Biosystems Science a
Kilde 16 ↗
Originalcitat
nd Engineering D-CHAB Chemistry and Applied Biosciences D-EAPS Earth and Planetary Sciences D-GESS Humanities, Social and Political Sciences D-HEST Health Sciences and Technology D-INFK Computer Science D-ITET Information Technology and Electrical Engineering D-MATH Mathematics D-MATL Department of Materials D-MAVT Mechanical and Process Engineering D-MTEC Management, Technology and Economics D-PHYS Physics D-USYS Environmental Systems Science Table of contents and legal Sitemap Imprint Accessibility Statement Disclaimer & Copyright Data protection © 2026 Eidgenössische Technische Hochschule Zürich JavaScript has been disabled in your browser
EthelUniversitets-hostet tjenesteDer gælder betingelser
Intern lokal ETH-infrastruktur
Ethel · Adgangsbetingelser
Undervisere og kursusdeltagere i deltagende ETH-kurser (kontakt Gerd Kortemeyer). Data: ukendt Vurdering: ukendt
Intern lokal ETH-infrastruktur
ukendt
Ethel · Officiel dokumentation
Kilde 1 ↗
Originalcitat
AI Tools at ETH Zurich × Contact Deutsch Applications Current: AI Tools at ETH Zurich AI Tools at ETH Zurich This page is only available in English. Safe use of AI Before exploring the capabilities of these tools, follow our Information Security Guidelines to ensure compliance with ETH Zurich’s data protection and security standards. Review the official recommendations here . Users are responsible for using AI tools in an authorized and responsible manner. See Guide for Users of AI at ETH Zurich . Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential
Kilde 2 ↗
Originalcitat
data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted. AI‑generated output must be reviewed before use. Data classification Information owners classify the information according to the risk-oriented approach as explained in RSETHZ 203.28 Appendix 1a Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may
Kilde 3 ↗
Originalcitat
only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored. Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”. Information is considered public if it is authorized for publication by the responsible body. Classification recommendation: AI
Kilde 4 ↗
Originalcitat
tools managed by IT Services The tools in this section are enabled and managed by ID CCR Cloud Service Center. Costs AI tools included in personal cloud subscriptions, such as Microsoft 365 and Google Workspace, are available free of charge. Any additional AI tools not included in these subscriptions are at the user’s own expense. Microsoft AI tools Tool Data residency & processing location Cloud assessment Access & license Copilot Free GLOBAL NO ASSESSMENTS No sign-ins M365 Copilot Chat CH/EU Approved for internal data ETH members | M365 subscription M365 Copilot CH/EU The work mode is approved for CONFIDENTIAL data when the web search option is turned off. In the work mode , web search is
Kilde 5 ↗
Originalcitat
disabled by default. The web mode is approved for INTERNAL data. ETH employees | M365 subscription + M365 Copilot license Microsoft Copilot Studio CH/EU NO ASSESSMENTS ETH employees | M365 subscription + Copilot Studio license Azure Foundry Selected location NO ASSESSMENTS ETH employees | Azure subscription Enterprise Data Protection , i.e., data is not used in model training and not reviewed by human reviewers. A cloud assessment is in place. Google AI tools Prerequisite: Google Workspace subscription Google AI Tools have not yet been assessed by the CISO. Tool Data residency & processing location Access & License Gemini Chat EUROPE ETH members | Google Workspace subscription Google AI Pro
Kilde 6 ↗
Originalcitat
for Education EUROPE ETH employees | Google Workspace subscription + Google AI Pro license Prerequisite Only employees with a Google Workspace subscription can order a Google AI Pro license . NotebookLM GLOBAL ETH members | Google Workspace subscription Google AI Studio GLOBAL ETH members | Google Workspace subscription Restricted access ETH members with a Google Workspace subscription can request to participate the Google AI Studio Pilot by contacting servicedesk@id.ethz.ch Conversational Agents Selected location ETH members | Google Workspace subscription + Google Project Google Vertex AI Selected location ETH employees | Google Workspace subscription + Google Project Enterprise Data Prote
Kilde 7 ↗
Originalcitat
ction , i.e., data is not used in model training and not reviewed by human reviewers. AI tools provided by ETH Library AI-powered tools and services by the ETH Library Unsupported tools The tools in this section are not managed by ID CCR Cloud Service Center. As far as we know, NO cloud assessments are in place. The use of these tools are on user's own responsibility. Tool Data location OpenAI ChatGPT US DeepSeek CHINA Grok US Perplexity Focuses on search engine and text generation providing accurate results with references US Claude A helpful, honest, and harmless assistant relying on the training data until 04/2024. No search function US SwissAI Apertus GLOBAL Informaniak Euria CH CamoCopy
Kilde 8 ↗
Originalcitat
GDPR-compliant AI tools EU Custom GPT Data Analyst for complex data analysis and visualization US/EU/CH Coding assistance GitHub copilot US/EU Cursor AI US Google Antigravity US/EU/CH On-prem AI tools Tool Data location Contact Ethel CH Gerd Kortemeyer, Rectorate and AI Center, ETH Zurich, kgerd@ethz.ch Guides Safe use of AI Guidelines in specific areas × Copyright © 2026 Informatikdienste der ETH Zürich • Powered by Scroll Sites and Atlassian Confluence
Microsoft Copilot Free (Web)Ikke godkendt af universitetetIkke anbefalet
Ingen login
Microsoft Copilot Free (Web) · Adgangsbetingelser
Offentlige / uautoriserede brugere. Data: Kun offentlige data; strengt forbudt for alle ETH-data. Vurdering: ukendt (ikke-understøttet eksternt værktøj; anbefales ikke til akademisk arbejde; underlagt kursusregler)
Ingen login
Gratis
Microsoft Copilot Free (Web) · Officiel dokumentation
Kilde 1 ↗
Originalcitat
The service can be accessed in a protected environment via a Microsoft 365 Cloud Subscription , which is provided by ETH Zürich.
Kilde 2 ↗
Originalcitat
The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data .
Kilde 3 ↗
Originalcitat
Uploading strictly confidential data to external cloud services, including AI tools, is prohibited. The work mode of M365 Copilot without the web search option is approved for use with confidential data . The web search option is disabled in the work mode by default. M365 Copilot Chat and the web mode of M365 Copilot is approved for internal data . AI tools lacking a cloud assessment have not been released for use with ETH data. Processing confidential, strictly confidential, or protected personal data is not permitted.
Kilde 4 ↗
Originalcitat
Information is classified as “ strictly confidential ” if unauthorised access could damage the interests of ETH Zurich. It must not be stored on external cloud services and may only be shared with a specifically authorised group under strict security precautions. Information is classified as “ confidential ” if unauthorised access could significantly damage the interests of ETH Zurich. It must be limited to specific authorised individuals and securely stored.
Kilde 5 ↗
Originalcitat
Information is classified as “ internal ” if unauthorised access could damage the interests of ETH Zurich. Such information is therefore generally intended solely for members of ETH Zurich. Unless explicitly stated otherwise, all data and information at ETH is considered “internal”.
Kilde 6 ↗
Originalcitat
How Can I Use Microsoft Copilot? × Contact Deutsch Microsoft 365 Manuals and Documentation Guides - Copilot Current: How Can I Use Microsoft Copilot? How Can I Use Microsoft Copilot? This page is only available in English. The Enterprise Plan for Microsoft 365 Copilot has the two following plans: Microsoft 365 Copilot Chat (with Enterprise Protection) available to all ETH users with a Microsoft 365 Subscription Microsoft 365 Copilot (available only for ETH employees) requires a basic license ( Microsoft 365 A5 for faculty ) plus an add-on license that can be acquired through the IT-Shop . Further detailed information can be found here: Microsoft Copilot Manual Navigate to https://copilot.mic
Kilde 7 ↗
Originalcitat
rosoft.com/ Sign in with your ETH Account (username@ ethz.ch ) and credentials. After a successful login, you can use Microsoft Copilot. Getting started with Copilot on Windows - Microsoft Support × Copyright © 2026 Informatikdienste der ETH Zürich • Powered by Scroll Sites and Atlassian Confluence

Opdateringer af posterSeneste 90 dage · Registrerede opdateringer: 2

Opdateringer af denne sides gennemgåede registrering, ikke en bekræftelse af, at universitetet har ændret sin politik.

2026-10-27Sideopdatering
Andre poster · Sideopdatering
Vis revision
Indlæser revision…
2026-10-14Registreringstilføjelse · Rettelse af websted
Akademisk integritet · Undervisning · 4 flere emner · Records involved: 15 (11 corrected / adjusted · 4 added)
Vis revision
Indlæser revision…

Gennemgåede påstande

15

Officielle kilder

10 officielle webadresser; 10 kildeangivelser og deres gemte versioner er bevaret.

AI Tools & Licenses | ETH Zurich

ethz.ch

Officiel universitetskilde

Hash og indfanget version

AI Tools & Licenses | ETH Zurich · 2026-10-12T06:11:27.987897Z

Representation unspecified in this historical record; do not assume an original-file hash.

0a7517b303b91ff263accfe492e8941ddfda122b172c5f465392b2bebb5afb1d

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

ETH IT Services: Google Gemini & Google AI Pro for Education

unlimited.ethz.ch

Officiel universitetskilde

Hash og indfanget version

ETH IT Services: Google Gemini & Google AI Pro for Education · 2026-10-12T06:11:38.470205Z

Representation unspecified in this historical record; do not assume an original-file hash.

ecd7b331c95826fa240b55867a20b122de6b2b43bcbb61bbe9152a1e9e2422d9

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

ETH IT Services: AI Tools at ETH Zurich & Security/Data Classification

unlimited.ethz.ch

Officiel universitetskilde

Hash og indfanget version

ETH IT Services: AI Tools at ETH Zurich & Security/Data Classification · 2026-10-12T06:11:37.020618Z

Representation unspecified in this historical record; do not assume an original-file hash.

36be2d664bee7d73dc98b2a88fe13b98ea4a029c161a67569491550deb6e402d

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

AI in Teaching and Learning — ETH Zurich

ethz.ch

Officiel universitetskilde

Hash og indfanget version

AI in Teaching and Learning — ETH Zurich · 2026-10-12T06:11:25.974620Z

Representation unspecified in this historical record; do not assume an original-file hash.

05e768573acf7a8f060bb9736d5648b54b290410bc6dcf16621d24d2959c970b

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

Generative AI in Teaching and Learning - Guidelines (December 2024)

ethz.ch

Officiel universitetskilde

Hash og indfanget version

Generative AI in Teaching and Learning - Guidelines (December 2024) · 2026-10-12T06:11:24.710720Z

Representation unspecified in this historical record; do not assume an original-file hash.

b883ba3727ad5e0f8af3d12a1846e3a6ecdab325e9ab95b75f80422463cc66ca

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

ETH Library: Plagiarism and Generative Artificial Intelligence

library.ethz.ch

Officiel universitetskilde

Hash og indfanget version

ETH Library: Plagiarism and Generative Artificial Intelligence · 2026-10-12T06:11:33.023336Z

Representation unspecified in this historical record; do not assume an original-file hash.

79bad72c13cb4c2710053173b3ed9ce7ca1211954827ca5aa11aed36446c3c02

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

Academic Integrity — AI in Education

ethz.ch

Officiel universitetskilde

Hash og indfanget version

Academic Integrity — AI in Education · 2026-10-12T06:11:26.636240Z

Representation unspecified in this historical record; do not assume an original-file hash.

7cea3253df7b975f75dc10a21d31010df7046af69f82016d9ebd58656a8eeebc

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

ETH Zurich Declaration of Originality (Bachelor & Master)

ethz.ch

Officiel universitetskilde

Hash og indfanget version

ETH Zurich Declaration of Originality (Bachelor & Master) · 2026-10-12T06:11:28.941491Z

Representation unspecified in this historical record; do not assume an original-file hash.

d3a3ed4457d6c6f5a7dea2d2e084ff201b1f63d8d56af61927e536cb5897f23a

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

ETH Zurich Doctoral Examination Registration & Authorship Declaration

ethz.ch

Officiel universitetskilde

Hash og indfanget version

ETH Zurich Doctoral Examination Registration & Authorship Declaration · 2026-10-12T06:11:31.841193Z

Representation unspecified in this historical record; do not assume an original-file hash.

2fc181ae19d4630b43464d380b8aa66da7baf4df996ef3b20b3e2ee8db373937

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

ETH Library: Plagiarism Check & Turnitin Originality

library.ethz.ch

Officiel universitetskilde

Hash og indfanget version

ETH Library: Plagiarism Check & Turnitin Originality · 2026-10-12T06:11:34.938479Z

Representation unspecified in this historical record; do not assume an original-file hash.

fe42a443c6f65d498f8cd583c4947e8ca5354d1cf5a292e6fd721c9ecd20a0e3

Hashes identify a captured version; they do not establish current policy or review approval.

Officiel kilde ↗

Postoplysninger, JSON og citering

Review layers: claim review applies to the claims only. Analysis is a separate machine-generated summary awaiting review. Coverage scores do not establish complete policy knowledge.

Claim review state: agent_reviewed
Analysis review state: machine_candidate
Record revision: v6-audit-fix-20261014
Dataset release: public-release-20260924-002
Correction published: 2026-10-14T02:23:21.633198Z
Student snapshot: Not published

Current reviewed JSON · Historical baseline

Published scoped record correction; current JSON and MCP use the same reviewed record, while the previous baseline is retained in history. Institution-wide audit remains PARTIAL.

Det offentlige endepunkt for øjebliksbilleder af studerende returnerer HTTP 404 (ikke fundet). Et lokalt udkast til kandidat med 6-dimensioner er udarbejdet i outputs/student-topics.json, men er ikke et publiceret, stærkt øjebliksbillede af studerende.
Google AI-værktøjer (Gemini Chat, NotebookLM, Google AI Pro) har ikke gennemført CISO-cloudvurderingen (status OPEN). Ikke-vurderede værktøjer frigives ikke til ETH-data, der er fortrolige, strengt fortrolige eller beskyttede personoplysninger.
Adgang til Turnitin Originality er begrænset til professorer og undervisere, men betalingsstrukturen (gratis/central vs. institutbaseret) fremgår ikke af den officielle side.
Disciplinære ankeinstanser, juridiske tidsfrister og forvaltningsdomstolsprøvelser styres af almindelig schweizisk forvaltningsret og ETH-regler, men fulde lovtekster blev ikke crawlet.
Eksplicitte undtagelser vedrørende handicapkompensation eller kompenserende teknologi for generativ AI er ikke detaljeret beskrevet i de nuværende undervisningsretningslinjer.

Current reviewed JSON

Spørgsmål, som studerende stiller

Spørgsmål, som studerende stiller

Må jeg bruge AI til mit kursusarbejde?

ETH Zurich slår til lyd for en proaktiv og ansvarlig tilgang til generativ AI i opgaveforløb. Studerende bærer fortsat det fulde ansvar for indholdet af ethvert arbejde, de afleverer. Generativ AI kan spille en supplerende rolle (såsom idéudvikling, strukturering eller kodehjælp), men må aldrig erstatte den personlige indsats. Undervisere har myndighed til at bestemme, om og hvordan generativ AI må anvendes i deres kurser.

Vis evidens →

Skal jeg oplyse om brug af AI?

ETH Zürichs originalitetserklæring giver to muligheder: at bekræfte, at ingen generativ AI er blevet brugt, eller at bekræfte, at godkendt generativ AI er blevet brugt, og at angivelsen er aftalt med vejlederen. Bachelor- og masterafhandlinger kræver en underskrevet originalitetserklæring, mens krav til andre skriftlige opgaver afhænger af de gældende studieregler. Ph.d.-kandidater erklærer selvstændigt forfatterskab gennem tilmeldingsprocessen til doktorgradseksamen. Den tilladte brug, formatet for angivelse og eventuelle undtagelser skal følge kravene fra den relevante vejleder og enhed. ETH Zürichs undervisningsretningslinjer instruerer studerende i at være gennemsigtige om deres brug af generativ AI og i at angive, hvilke værktøjer der bidrog til hvilke dele af arbejdet, i overensstemmelse med gældende instruktioner fra underviseren og citationskonventioner. Ved videnskabelig skrivning og afhandlinger bør studerende på forhånd aftale med deres vejledere, hvilken AI-brug der er godkendt, og hvilke angivelseskrav der gælder. ETH Library-vejledningen anerkender en undtagelse, hvor en vejleder udtrykkeligt har accepteret, at en bestemt AI-brug ikke skal angives særskilt. Uden en sådan aftale kan ikke-angivet brug udgøre en overtrædelse af akademisk integritet.

Vis evidens →

Må jeg uploade følsomme data?

Upload af strengt fortrolige oplysninger til eksterne cloud-AI-tjenester er forbudt. Fortrolige oplysninger må kun behandles i specifikt evaluerede konfigurationer (såsom M365 Copilot Work-tilstand uden websøgning). M365 Copilot Chat er godkendt til interne data. Google AI-værktøjer er endnu ikke blevet vurderet af CISO (vurdering OPEN) og er ikke frigivet til ETH fortrolige, strengt fortrolige eller beskyttede personoplysninger. Ophavsret og databeskyttelse skal til enhver tid overholdes.

Vis evidens →

Hvilke værktøjer har jeg adgang til?

ETH Zürich anbefaler Microsoft Copilot, Google Gemini og Google NotebookLM til undervisningsformål. Chatversionerne i basisudgaven er inkluderet i institutionelle abonnementer på Microsoft 365 og Google Workspace, gratis for studerende og medarbejdere, og de tilbyder virksomhedsdatabeskyttelse, hvor brugerdata ikke bruges til modeltræning. Google-værktøjerne har endnu ikke gennemført CISO-vurderingen (ÅBEN), så produkttilgængelighed udgør ikke en godkendelse af dataklassificering. Værktøjets tilgængelighed betyder ikke tilladelse til bedømmelse; undervisere afgør tilladelser til bedømmelse.

Vis evidens →

Hvad er eksamensreglerne?

Præstationsvurderinger skal gennemføres selvstændigt og personligt af den studerende. Brug af uautoriserede hjælpemidler, herunder generative AI-værktøjer, som eksaminator ikke har tilladt, udgør akademisk snyd og kan medføre disciplinære foranstaltninger i henhold til ETH Zürichs bekendtgørelse om disciplinære foranstaltninger.

Vis evidens →

Hvilke regler gælder for forskning og publicering?

Forskere og studerende er fuldt ud ansvarlige for integriteten af deres videnskabelige arbejde og skal verificere AI-genereret output, referencer og fakta. Generative AI-værktøjer kan ikke angives som forfattere eller citeres som autoritative kilder. Specialer kræver underskrevne tro og love-erklæringer; ph.d.-kandidater erklærer uafhængigt forfatterskab ved tilmelding til ph.d.-forsvaret. Forskere skal overholde gældende retningslinjer fra bevillingsgivere (f.eks. SNSF) og forlag.

Vis evidens →

Hvad udgør akademisk forseelse?

ETH Library stiller Turnitin Originality til rådighed for berettigede professorer og lektorer til kontrol af tekstlighed, underlagt dataklassificering og vejledning om ansvarlig, selektiv brug. Kun offentlige eller interne dokumenter må uploades. Lighedsscreening er adskilt fra detektering af AI-forfatterskab; ETH Zurich advarer om, at teknisk detektering af AI-genereret indhold er upålidelig og ikke bør behandles som afgørende bevis. Overtrædelser af GenAI-retningslinjer såsom brug af uautoriserede hjælpemidler eller manglende oplysning om deres brug er underlagt disciplinære sanktioner i henhold til eksisterende regler for præstationsvurdering og tro- og loveerklæringen om originalitet.

Vis evidens →

Hvilken vejledning gælder for undervisning?

Undervisere opfordres til at tilpasse undervisnings- og bedømmelsesstrategier til GenAI, drøfte acceptabel brug med de studerende, tydeligt kommunikere, hvilke tilladelser der gælder i kurset, og fungere som rollemodeller ved at gøre brugen af GenAI synlig i undervisningsmaterialer og bedømmelsesforløb, samtidig med at de sikrer overholdelse af databeskyttelsesregler.

Vis evidens →

Hvilke sikkerheds- og indkøbskrav gælder?

ID CCR Cloud Service Center administrerer institutionelle cloud-AI-værktøjer. Værktøjer, der mangler cloud-vurderinger (herunder tredjepartsværktøjer såsom kommerciel ChatGPT, Claude, DeepSeek, Perplexity, Grok), anvendes på brugerens eget ansvar og må aldrig behandle fortrolige, strengt fortrolige eller beskyttede persondata. Institutionel tilkøbssoftware (såsom Google AI Pro eller M365 Copilot) kræver medarbejderlicensanskaffelse via IT-Shop; specifikke gebyr- og finansieringsordninger afhænger af gældende organisatoriske procedurer.

Vis evidens →

Hvornår blev denne optegnelse kontrolleret?

Registrering kontrolleret: 2026-10-12

Registreringsversion →

Hvilke officielle kilder understøtter disse regler?

Officiel kilde →

Hvor kan jeg læse hele dokumentationen?

15 · Fuldstændig politisk dokumentation

Fuldstændig politisk dokumentation →