Responsible AI
Responsible AI policy
This policy sets out the commitments we make about the AI features in every Oxonia product, for the universities that use them and their students and staff.
1. About this policy
Last updated: 10 October 2026.
This policy applies to the AI features in all Oxonia products and on our website. It forms part of the terms of service and sits alongside our AI principles, which explain the design ideas, and the AI fair use policy, which sets the rules for users. In our legal documents, "Oxonia Academy", "Oxonia", "we" and "us" mean Oxford City Academy for Education and Development (OAED), of which Oxonia Academy is part. Its registered office is [Registered office address].
2. People approve official records
AI drafts. People decide. Anything that becomes an official record waits for a named person with the right role to approve it. That includes marks in Oxonia Mark, minutes in Oxonia Senate, policies in Oxonia Policy and regulator returns in Oxonia Nexus. The approver can change or reject the draft, and the approval is recorded in the audit log. No setting turns this off. See How AI drafts are approved.
We do not use AI to make decisions about a person that have legal or similarly significant effects without a person deciding.
3. Being open about AI
- AI drafts are labelled as drafts until a person approves them.
- Answers point to their source, such as the minute in a lecture, the page in a thesis or the words in a script. See Citations and sources.
- Where AI is unsure, it says so. In Oxonia Mark, uncertain, illegible and borderline scripts go to a review queue for people.
- The AI fellow student in Oxonia Tutor is presented as AI, not as a real classmate.
4. Your data and AI training
- We do not use Customer Data, including students' work, recordings and exam scripts, to train or fine-tune AI models, ours or anyone else's, unless the university agrees in writing in its order form.
- The AI service providers we use are named on our sub-processors page. We use them under business terms that do not allow them to train their models on the data we send by default. They may keep that data for a limited period for abuse and safety monitoring, under those terms.
- A university can instead connect its own AI keys, in which case its own agreement with that provider applies. See Bring your own AI keys.
- We send each AI service only what it needs for the task.
5. Academic integrity
- Each university sets its own rules for AI in teaching and assessment. Oxonia Policy helps it approve an AI policy, set the level of AI use allowed for each module and keep a register of AI tools.
- Oxonia AI Hub teaches students to use AI well, explains the rules with the clause, and collects their AI-use declarations.
- Oxonia Tutor is designed to discuss a student's own work and challenge their reasoning, not to produce work for them to submit.
- Using Oxonia to break a university's assessment rules is against our acceptable use policy.
6. Fairness, accuracy and safety
- Students see why each mark was given, and appeals and re-marks are tracked.
- Any flag raised in a proctored online exam is a prompt for a person to review, not a finding of misconduct.
- We do not use AI to infer students' or staff members' emotions.
- We evaluate AI features before release, monitor reported problems, and fix or withdraw features that do not meet this policy.
- Features are designed for accessibility, including captions and support for several languages. See the accessibility statement.
7. Law and regulation
We design our AI features to help universities meet the data protection rules on automated decisions in the UK, the EU and the other countries we serve. Under the EU AI Act, some education uses, such as evaluating learning outcomes and monitoring students during tests, are classed as high-risk. Where the Act applies, we provide the information and human oversight features universities need for their own duties, and we work with them on their assessments.
8. Reporting a problem
If an AI feature produced something wrong, unfair or harmful, tell us at support@oxonia.ai. University staff can also log it in the incidents log in Oxonia Policy, and students should tell their lecturer or university. Questions about this policy: legal@oxonia.ai.
Related