Technical University of Munich

TUM ProLehre student guidance specifies that generative AI tools may be used for written coursework, term papers, essays, and theses only when AI tool use is explicitly permitted for that specific assessment by the instructor or module specifications. Where no statement regarding AI use is made in the assignment or module handbook, students are strongly advised to ask the instructor in advance; using AI without permission may be treated as attempted deception under APSO § 22 or comparable regulations. When authorized, transparent documentation of tool use and prompts is commonly required, and students retain full personal responsibility for submitted work.

In the APSO reading version as amended by the12th amending statute, §22(1) says attempting to influence an examination result through deception or unauthorized aids leads to a failing grade. Suitable software may check unacknowledged text or sources, while the examiner determines whether deception occurred. Under §22(2), disruption may lead to exclusion from that examination and a failing grade. In serious or repeated deception or disruption cases, the examination committee may exclude the student from further examinations in the degree program, with final failure of the degree examination (§22(3)). Under §24(6), an examination failed because of §22 deception or disruption may be retaken only once. These are general examination provisions, not a separate AI-specific sanction schedule.

The Guidelines for Writing Theses and Seminar Papers (Juli 2026) issued by the TUM Chair of Marketing and Consumer Research permit the use of generative AI tools (such as ChatGPT, Claude, Perplexity) as research aids for seminar, Bachelor's, and Master's theses, subject to mandatory declaration inserted before the bibliography. The declaration requires students to specify the exact tools used and the specific tasks performed (brainstorming, literature search, structuring, code generation, hypothesis generation, translation, image generation, or language improvement), and confirm that they personally reviewed the entire thesis and accept full responsibility. If no AI tools were used, a formal non-use statement must be included ('Ich habe keine generativen KI-Tools oder -Hilfsmittel verwendet.'). AI-generated images must cite the specific tool, and the use of AI for empirical data collection and analysis is prohibited for data protection reasons unless explicitly approved by the thesis supervisor. These are the reviewed chair guidelines; no broader TUM rule established.

TUM ProLehre guidance instructs students and instructors not to enter personal, sensitive, or confidential data into AI tools, including systems approved by TUM. Prohibited categories include names, matriculation numbers, email addresses, patient records, research participant data, exam data, and confidential research results. Users are instructed to anonymize content prior to entry, utilize generalized or neutral sample texts, and consult instructors or the TUM Data Protection Officer when uncertain.

TUM ProLehre student guidance specifies that teaching and study materials such as lecture slides, scripts, recordings, or past exams are generally protected intellectual property and may only be uploaded to or processed in AI tools with explicit instructor consent. This consent requirement applies even when materials are uploaded solely for personal study purposes, including summarization, transcription, or translation. Where AI tools are used to enhance accessibility of learning materials, such as transcribing a lecture recording due to an impairment, ProLehre recommends consulting the instructor or academic advising in advance.

The reviewed TUM-SGwP statute requires researchers working at TUM to follow recognized scientific standards, document work, question results and act honestly (§1). Supervisors must verifiably give the statute to students conducting scientific thesis work (§22). Under §8, primary and evaluated data and evaluation software used for publications must be kept on durable secure media in the originating institution; retention depends on discipline and is ordinarily ten years from the relevant publication, with reasons for non-retention or shorter periods documented. Authorship requires a genuine, traceable contribution and honorary authorship is prohibited (§12). Under §13, scientific misconduct involves knowingly or grossly negligently making false statements, violating others' intellectual property or impairing their research in a scientifically relevant context. Anyone may approach the ombudsperson concerning suspected misconduct (§16).

In the reviewed APSO reading version, §25(3) grants access to written examination work, examiner reports and oral-examination minutes on a reasoned written application after results are announced. The application must reach the examination committee chair within one month of announcement; the chair determines the time and place with the examiner. Under §21, examination-procedure defects must be raised promptly in writing to the committee chair or examiner. This inspection window does not establish a deadline for disciplinary or administrative appeals, which remain unreviewed.

M365 Copilot Chat is listed for Technical University of Munich in an official university AI tools source. Derived availability: allowed. Derived endorsement type: officially endorsed.

Grammarly is listed for Technical University of Munich in an official university AI tools source. Derived availability: allowed. Derived endorsement type: institutionally licensed or procured.

M365 Copilot Chat is listed for Technical University of Munich in an official university AI tools source. Derived availability: allowed. Derived endorsement type: institutionally licensed or procured.

OneTutor is listed for Technical University of Munich in an official university AI tools source. Derived availability: conditionally allowed. Derived endorsement type: officially endorsed.

Artemis is listed for Technical University of Munich in an official university AI tools source. Derived availability: conditionally allowed. Derived endorsement type: officially endorsed.

Iris is listed for Technical University of Munich in an official university AI tools source. Derived availability: conditionally allowed. Derived endorsement type: officially endorsed.