University of Oxford

Those setting summative assessment must - declare whether and how students can use AI in summative assessment, e.g. by a category system for different assignments on courses. - review their summative assessment design and criteria by task to ensure alignment with the permitted use of AI - where students are authorised to use AI tools for their summative assessments, ensure that there is an equality of baseline provision of appropriate AI tools. - specify the forms of declaration expected of students - only identify suspected unauthorised use of AI in summative assessment through the marking process or through AI detection tools that have university endorsement. NB. as at the date of the l...

Source language

English (en)

Snapshot hash

dc16c703b23186b5ab18fa8cf527cbe294fbe4b153872be7f2a0752a6f28805c

Retrieved

May 6, 2026

Evidence locator

Staff duties section

Students are required to include a statement on their use of Gen AI in their final submitted thesis. This is effective as of submission in Trinity Term 2026, but it is recommended that such a statement is included in every thesis submitted from the point of publication of this guidance. The statement should be placed immediately after the abstract. The statement must include a formal declaration that any Gen AI use complies with University, divisional and (where applicable) departmental guidance, where and how Gen AI has been used in preparation of the thesis and summarising how specific uses of Gen AI will be referenced in the text

Source language

English (en)

Snapshot hash

6919aedf33483a2f0ea58ff0ecf690348f7d294d0d20bba8049f6e53dea1fa06

Retrieved

May 6, 2026

Evidence locator

Acknowledging and referencing use of generative AI tools section

Students undertaking summative assessment must - complete summative assessment in line with the declaration as to whether and how AI can be used in each specific assignment they will complete for their course. - acknowledge their use of AI as part of the summative assessment submission and use a formal declaration in the format prescribed by the assessment setter. - be aware that submitting work that breaches the specifications defined for a particular assignment constitutes cheating and may constitute plagiarism; cases of suspected unauthorised use of AI will be handled under the usual disciplinary Regulations and using the associated processes.

Source language

English (en)

Snapshot hash

dc16c703b23186b5ab18fa8cf527cbe294fbe4b153872be7f2a0752a6f28805c

Retrieved

May 6, 2026

Evidence locator

Student duties section

The policy is based on three principles for acceptable use of AI which were endorsed by Education Committee during Trinity term 2025: - Educational practice in teaching and assessment must be grounded in values of integrity, honesty and transparency. These values need to be clearly articulated and frequently discussed. - Every discrete unit of assessment must be carefully designed to be fit for its specific purposes. These purposes need to be clearly articulated to students. - Every element of summative assessment must be accompanied by a clear explanation of what appropriate assistance is permitted and what is forbidden. Where assistance is permitted, the assignment should specify exactl...

Source language

English (en)

Snapshot hash

dc16c703b23186b5ab18fa8cf527cbe294fbe4b153872be7f2a0752a6f28805c

Retrieved

May 6, 2026

Evidence locator

Three principles section

Before any processing of Internal or Confidential information using generative AI services, the following steps must be taken to mitigate risk. 1. As with all service providers holding or processing university information, information supplied to the tool in the form of questions or other artefacts is typically stored by the third-party service provider and is subject to the threats from cyber criminals and other malicious actors, such as hostile nation states. Therefore, all cloud-based Generative AI tools should be subject to a security risk assessment before being used. The Information Security GRC Team has a TPSA tool to help complete an assessment. It is generally not possible to com...

Source language

English (en)

Snapshot hash

9bbf71bffc824494cb7a1d67b222766e18ac06caeb82946a797b485c0561c17e

Retrieved

May 6, 2026

Evidence locator

Risk to confidentiality section

ChatGPT Edu, licensed via the AI Competency Centre, has been approved for processing of Confidential University data by the Information Security team. University data processed by ChatGPT Edu under an AI Competency Centre licence will not be used to train the AI model.

Source language

English (en)

Snapshot hash

9bbf71bffc824494cb7a1d67b222766e18ac06caeb82946a797b485c0561c17e

Retrieved

May 6, 2026

Evidence locator

ChatGPT Edu section

Google Gemini, licensed via the AI Competency Centre, has been approved for processing of Confidential University data by the Information Security team. University data processed by Google Gemini under an AI Competency Centre licence will not be used to train the AI model.

Source language

English (en)

Snapshot hash

9bbf71bffc824494cb7a1d67b222766e18ac06caeb82946a797b485c0561c17e

Retrieved

May 6, 2026

Evidence locator

Google Gemini section

The University provides several enterprise-grade AI tools that have passed internal Third-Party Security Assessments (TPSA). Accessing these tools via Single Sign-On (SSO) ensures that user data is not used to train external AI models. Confidential data must only be used with the University's approved, SSO-protected platforms.

Source language

English (en)

Snapshot hash

6abb77311f43f70e8af1d23ccb75bbd35bd34098129bb0260b38de9d4effb1f2

Retrieved

May 6, 2026

Evidence locator

University-supported AI Platforms section

Substantive original writing by Gen AI, including either verbatim or closely paraphrased use of Gen AI content, for, e.g., chapters, or parts of chapters, including introduction or conclusion chapters or for a literature review, would fall under the definition of plagiarism or be otherwise a failure of research integrity and is therefore not permissible. The use of generative AI to produce plots or data visualisations directly from prompts is prohibited. Private or confidential data must not be entered into third-party AI tools.

Source language

English (en)

Snapshot hash

6919aedf33483a2f0ea58ff0ecf690348f7d294d0d20bba8049f6e53dea1fa06

Retrieved

May 6, 2026

Evidence locator

Non-Permissible use section

External custom GPTs should not be used to process confidential University data or sensitive personal data. If you are considering inputting internal University data to an external custom GPT, please discuss this with the Information Security GRC team in advance. Any inputting of personal data to an external custom GPT should also be discussed with the Information Compliance team.

Source language

English (en)

Snapshot hash

9bbf71bffc824494cb7a1d67b222766e18ac06caeb82946a797b485c0561c17e

Retrieved

May 6, 2026

Evidence locator

Accessing external custom GPTs section

Be aware that any information incorporated in a custom GPT, either as Instructions specifying the behaviour of the GPT or as Documents uploaded to the GPT, may be accessed by users of the custom GPT. No non-public University data should be incorporated in any custom GPT that you intend to share with external users. For the avoidance of doubt, this includes any confidential or internal University data, as well as any personal data.

Source language

English (en)

Snapshot hash

9bbf71bffc824494cb7a1d67b222766e18ac06caeb82946a797b485c0561c17e

Retrieved

May 6, 2026

Evidence locator

Sharing custom GPTs externally section

There should be no use of unapproved AI transcription bots in Teams meetings by any participants. It is permissible to record meetings using the inbuilt Teams Transcription facility or Microsoft's Copilot subject to appropriate data protection considerations. Use of any AI transcription bot services other than the inbuilt Teams Transcription or Microsoft's Copilot should be avoided. Meeting options should be set by the organisers so as to prevent internal or external meeting participants from adding unapproved transcription bots.

Source language

English (en)

Snapshot hash

9bbf71bffc824494cb7a1d67b222766e18ac06caeb82946a797b485c0561c17e

Retrieved

May 6, 2026

Evidence locator

AI transcription bots in meetings section

The use of local editing tools—such as grammar assistants, code debuggers, and spell-checkers—is permitted and need not be declared. These tools only make small, local changes (for example, fixing spelling, grammar, or small pieces of code), usually affecting just a few words or tokens at a time. The use of AI tools for background research, language translation, creation of bibliography indices and general subject understanding is allowed and does not have to be declared. Use of Gen AI for coding purposes is permitted, where the coding serves a purpose in the research but is not the substantive output of the project.

Source language

English (en)

Snapshot hash

6919aedf33483a2f0ea58ff0ecf690348f7d294d0d20bba8049f6e53dea1fa06

Retrieved

May 6, 2026

Evidence locator

Permissible use section

Students using AI during their studies must learn and practise the same academic skills of note-taking and clear attribution which are safeguards against plagiarism, ensuring clear differentiation of their own work from any text or material derived from generative AI tools. Unauthorised use of AI falls under the plagiarism regulations and would be subject to academic penalties in summative assessments. Where the use of generative AI in preparing work for examination has been authorised by the department, faculty or programme, students should give clear acknowledgment of how it has been used in their work.

Source language

English (en)

Snapshot hash

9414bd7307cd58218608ce57b4e8bc2f6f1dd910fb69e1747ffcbc21fb7ca0d9

Retrieved

May 6, 2026

Evidence locator

Generative AI tools and plagiarism section