Policy on Artificial Intelligence

Most Nigerian lecturers and students already use AI tools, but few have been trained and most institutions lack rules for their use. The National Universities Commission (NUC) requires every university to adopt an institutional AI policy, constitute an AI Governance Committee and define acceptable and prohibited uses of AI in academic work. This policy is Babcock University's response, grounded in its Seventh-day Adventist mission and its motto, Knowledge, Truth, Service.

Three instruments at the centre

• A five-level scale telling students exactly how much AI each assessment allows
• A classification of University information telling everyone which data may be given to which AI tools
• A risk classification deciding how closely each institutional AI system is reviewed before use

Around these, the policy assigns duties to students, academic staff, administrative staff and researchers, protects personal data under the Nigeria Data Protection Act 2023, requires a human decision-maker for every decision that significantly affects a person, and provides a fair process for suspected misconduct. A separate Implementation Framework sets out the Committee structure, procedures, timelines, training and monitoring indicators. The policy is reviewed at intervals of not more than 12 months.

Abridged; the full definitions list is in the approved policy.

In national readiness surveys reported by the NUC, 95% of 1,202 lecturers and 94% of 519 students reported using AI tools, yet 65% of the lecturers had never received formal AI training and only 21% of students had been trained through their universities. Where use outpaces guidance, expectations conflict, integrity decisions are uneven, personal data leave the University through public AI services, and fabricated references enter scholarly work.

The NUC Framework directs each university to put AI governance in place, constitute an AI Governance Committee and adopt a code of conduct on acceptable and prohibited uses of AI in academic work. The NUC Committee Report recommends adopting a policy within 12 months and reviewing it at least annually.

Mission, motto and values

Babcock University is owned and operated by the Seventh-day Adventist Church. Its motto frames the policy: Knowledge — AI should deepen learning and research; Truth — AI output must never be passed off as something it is not; Service — AI should help the University serve students, patients, the Church and society, without weakening the human relationships that service depends on.

Policy section 1.

Objectives

• State, in one University-wide instrument, how AI may and may not be used at Babcock
• Let students and staff benefit from AI in studies, teaching, research and administration
• Keep assessed work, theses and publications honest and original
• Keep personal data and confidential information within the protection of Nigerian law
• Make a named person accountable for every institutional AI system
• Meet NUC and other regulatory requirements, build critical AI knowledge, and ensure AI narrows rather than widens gaps in access and outcomes

Persons and units covered

Students on every programme (undergraduate, postgraduate, sub-degree, part-time, distance or exchange); academic staff including librarians and adjunct, visiting and part-time lecturers; administrative, technical and support staff; research fellows, assistants and visiting scholars; every school, college, department, directorate, centre and unit including the Babcock University Teaching Hospital where it acts for the University in teaching and research; and contractors, consultants, vendors and partners to the extent their agreements and applicable law allow.

Technologies and activities covered

All AI technology, including generative AI, image, audio and video generators, machine learning, speech and language tools, predictive analytics, tutoring systems, agentic AI and AI features built into other software, in coursework, examinations, research, publication, administration and communication, on University systems or personal devices and accounts.

Exclusions and special safeguards

Purely personal AI use involving no University activity, information or reputation is outside the policy, though using a personal device or account never permits giving University information to an AI service. Where the University runs a school or programme for persons under 18, stricter safeguards set by the AI Governance Committee apply.

Policy section 2.

• Data protection: the Nigeria Data Protection Act 2023 governs AI processing of personal data. The GAID 2025, effective 19 September 2025, requires a DPIA for educational services involving student records, and documented safeguards, assessment of disparate outcomes and controlled testing before deploying AI.
• National frameworks: the policy gives effect at Babcock to the NUC Framework and NUC Committee Report (including the Framework for Ethical Usage of AI in Postgraduate Research), and is consistent with the National Artificial Intelligence Strategy 2025–2029.
• Copyright and cybercrime: the Copyright Act 2022 and the Cybercrimes (Prohibition, Prevention, etc.) Act 2015, as amended in 2024, apply to AI use as to any other activity.
• Research ethics: AI does not change the duty to obtain ethics approval under the National Code of Health Research Ethics.
• International: UNESCO's 2021 Recommendation on the Ethics of AI, its 2023 generative AI guidance and 2024 competency frameworks, and the 2023 COPE position statement on authorship and AI tools.
• Faith and context: AI use shall be consistent with the University's Christian ethical commitments and respect Nigerian languages, cultures and development priorities.

Nigeria had not enacted a standalone AI law at the date of approval; the policy will be amended when one takes effect. Where a law, regulation, funder or professional body imposes a stricter rule, the stricter rule applies.

Policy section 3.

Principles 1–8 are those of the NUC Framework; principles 9–13 reflect the University's own mission and academic tradition.

Policy section 4.

Every assessment in every course carries one of five AI use levels, stated in the course outline and the assessment brief. The scale is adapted from the published AI Assessment Scale (AIAS).

• Unless the course outline says otherwise, Level 2 applies to coursework and Level 0 to invigilated examinations, tests and practicals. A student in doubt asks the course instructor first.
• A student who keeps within the stated level and discloses as required is not penalised for using AI.
• Heads of Department keep levels consistent across comparable assessments in a programme.
• The levels govern students' assessed work only; staff use is governed by sections 8 to 10.

Policy section 5.

6.1 Information classes

Confidential and Restricted information shall not be entered into personal, free or unapproved AI accounts. The rule protects information the University holds about others; it does not stop a student using AI on his or her own work within the applicable level, or staff and researchers using AI on their own unpublished writing, analysis or code, unless it contains personal data, others' confidential material or content restricted by a funder, sponsor, partner or co-author.

6.2 Registers

The AI Governance Committee keeps and publishes a register of approved AI tools (permitted purposes, highest information class, conditions) and a register of institutional AI systems, each with a named business owner and technical owner. Until the first register is published, any widely used tool from a reputable provider is treated as approved for Public and Internal information only.

6.3 Risk classes

6.4 Contracts and autonomous agents

As far as the market allows, AI contracts bar vendor training on University information without consent, state where data are held, give audit and log access, require prompt breach notice, provide for deletion at exit and weigh energy and environmental impact. Autonomous agents shall not be connected to University accounts until approved as High risk, with the narrowest permissions, human confirmation before any consequential action, and safeguards against manipulation such as prompt injection.

Policy section 6.

Permitted uses for learning

Within the rules of each assessment and with honest disclosure, students may use AI to:

• Have concepts explained and generate practice questions for self-testing
• Brainstorm and outline, while doing the substantive thinking and writing themselves
• Find search terms and summarise readings, provided they locate and read every source before citing it
• Check the grammar, spelling and clarity of their own drafts, where the level permits
• Understand coding errors and examples, provided they can explain every line they submit
• Support learning as an assistive technology for disability, arranged through disability support services

Restrictions

Students shall not present AI-generated work as their own where own work is required; use AI in any Level 0 assessment; exceed the permitted level even with disclosure; invent or falsify data, results, quotations or references; impersonate another person or perform an oral or practical assessment in their place; or place examination questions, others' work or personal data, unpublished research or confidential material in tools not approved for it.

Disclosure, records and citation

• Use at Level 2 or above is disclosed in the form the level requires, using the AI disclosure form (Appendix A, Part 2) where full disclosure is needed
• Keep drafts, notes and AI interaction records until the result is final and any appeal period has ended
• Cite AI tools in the style the course requires (APA guidance in Appendix A, Part 3)
• Students answer for the accuracy, originality, citations and reasoning of everything they submit, including AI errors or invented references
• Every undergraduate final-year project, long essay or dissertation carries a signed declaration of compliance naming any AI tools used, countersigned by the supervisor

Access to tools

Where Level 3 or 4 is permitted or required, every student has access to an approved tool at no extra cost. No course may require a paid tool the University does not provide.

Policy section 7.

Teaching materials and curriculum

Academic staff may use approved AI tools to prepare teaching materials, examples, case studies, practice questions and rubrics, to draft assessments and marking schemes, and to translate or adapt content, and must check AI-produced materials for accuracy, bias and copyright. Only the finalised examination paper is Restricted. AI literacy and ethics form part of the General Studies curriculum for all undergraduates, and every department reviews its curricula to include relevant AI knowledge and skills.

Setting and marking assessments

• Course outlines state the AI use level of each assessment (Appendix A, Part 1), explained at the start of the course
• Assessments are designed to stay valid where AI is widely available: supervised work, oral defence, staged submissions, reflection on process, tasks tied to local data
• AI may score objective items; for interpretive or extended work it may only suggest draft marks, and the examiner reviews every mark and comment and decides the final mark. Students are told when AI is used in marking and may seek human review
• A detection score may be noted as one indicator but never alone supports an allegation, penalty or reduction of marks
• AI-based remote proctoring requires High-risk approval, advance notice, a non-AI alternative and human review of every flag

Postgraduate supervision

Supervisors discuss AI use with each student, review it at every progress stage and countersign the Appendix B checklist. Supervision itself is never delegated to AI.

Restrictions

Academic staff shall not enter student work, marks or personal data into unapproved tools; leave final marks, progression or disciplinary outcomes to AI; penalise a student whose disclosed AI use stayed within the level; treat a detection score alone as proof; use AI to invent or falsify research data, results or citations; or encourage or tolerate student AI use that breaches the policy.

Policy section 8.

• Office work: approved tools may draft letters and reports, summarise, format and translate documents, and analyse non-personal operational data. Staff are accountable for every document they issue.
• Chatbots: must say they are AI, offer a route to a human officer, and never give binding decisions or clinical advice.
• Decisions affecting individuals: no decision on admission, progression, grades, scholarships, discipline, appointment, promotion, health care or fees is taken solely by automated processing. A qualified officer reviews, may override and stays accountable; the person is told AI supported the decision, given an explanation of its role and offered human review.
• Meetings and records: participants are told before an AI note-taker is used. Confidential deliberations of Council, Senate, committees and disciplinary panels are not recorded or summarised with AI unless the tool is approved for Restricted information and the chair authorises it. AI minutes remain drafts until adopted, and AI is not used to create images or recordings of real officers, staff or students for official use without documented consent.
• Clinical and health services: AI, including at the Babcock University Teaching Hospital, supports and never replaces a licensed clinician and never processes patient information in a tool not approved for Restricted information.

Restrictions

Administrative staff shall not place personal, financial or other Confidential or Restricted records in unapproved tools; leave to AI any decision needing professional judgement; enter passwords or access keys in AI tools; use AI to get round procedures or controls; issue impersonal or misleading AI-generated communications; or install AI tools the ICT unit has not approved.

Policy section 9.

• Scope: AI may be used for literature searching and synthesis, study design, coding, analysis, modelling, transcription, translation and language editing. The researcher validates every result and checks every source.
• Documentation: record material AI use (tool and version, date, purpose, how outputs were checked, prompts or settings where they shaped a result) and disclose it in the methods section or an equivalent statement. Citing a source not read, or one that does not exist, is research misconduct whether or not AI produced it, as is fabricating or falsifying data, images, results or references.
• Postgraduate research: every proposal, dissertation and thesis contains an AI Usage and Compliance statement and the Appendix B checklist, signed by student and supervisor (including a "no AI used" statement where applicable). Interpretations, conclusions and the contribution to knowledge must be the student's own, and examiners may question the candidate on AI use.
• Human participants: ethics applications declare AI use involving participants, personal or health data. Participants are told when AI interacts with them or processes their data, and personal data go only into tools approved for them, with a lawful basis and ethics approval, de-identified wherever possible.
• Authorship: no AI tool may be named as author, co-author or co-investigator. Researchers follow the AI rules of journals, publishers, conferences and funders. Questions about AI use in published, non-retracted work are for the journal; the University opens an inquiry into published work only on a journal, publisher or funder request, a retraction or expression of concern for integrity reasons, or where the work is relied on in a University process and verifiable grounds exist. Screening under section 13.3 applies first.
• Confidential review and examination: manuscripts, grant proposals, theses and other material received in confidence are not uploaded to AI tools (work submitted for assessment may be checked with an approved University tool). Reviewers and examiners form and write their own judgements.
• Building AI systems: apply fairness, accountability and transparency across the life cycle: documented data sources, testing on representative data, access controls and monitoring for bias, drift and security weaknesses.

Policy section 10.

• The University maintains a designated Data Protection Officer, as the Nigeria Data Protection Act 2023 requires
• A DPIA is completed before deploying any AI system that processes student records, supports decisions about individuals, processes sensitive or children's data, involves systematic monitoring, or sends personal data outside Nigeria. It assesses the risk of disparate outcomes across groups and records controlled testing before deployment
• Only the personal data needed for the purpose are processed, and data are de-identified before AI processing wherever the purpose allows. Personal data leave Nigeria only on a basis the Act permits
• Data subjects are told when AI processes their personal data and their rights under the Act are honoured
• Approved AI systems have access controls, encryption and logging so operation can be traced and audited
• A suspected AI-related breach is reported at once to the Data Protection Officer and the ICT unit; the Nigeria Data Protection Commission is notified within 72 hours of the University becoming aware of a breach likely to risk individuals' rights and freedoms

Policy section 11.

No person to whom the policy applies shall use AI to:

1. Complete or assist with an examination, test or assessment in breach of its AI use level, or impersonate another person in any assessment
2. Fabricate or falsify data, results, images, references, quotations, transcripts, signatures or records
3. Produce deepfakes or synthetic media of a real person without consent, or for deception, harassment, defamation, pornography or political manipulation
4. Harass, bully, threaten, intimidate or discriminate against any person, or generate hateful content
5. Commit or facilitate fraud, phishing, impersonation, unauthorised access or any offence under the Cybercrimes Act
6. Decide solely by automated means about a person where the decision has legal or similarly significant effects
7. Conduct covert surveillance of students, staff or patients, or use emotion recognition or biometric categorisation to infer sensitive characteristics such as religion, ethnicity, health or political opinion
8. Place Confidential or Restricted information in unapproved tools, or bypass security controls
9. Infringe copyright, trade marks, patents or licence terms
10. Misrepresent the official position of the University or its proprietor, or use its name, logo or seal without authority
11. Upload material received in confidence for review or examination to AI tools, other than as section 10.6 allows
12. Develop or deploy AI systems that pose unacceptable risks or cannot comply with human rights obligations

Research and teaching that study the uses in items 3, 7 and 12 (such as detecting deepfakes or testing AI security) are permitted with ethics committee approval and safeguards preventing harm to real persons. Any violation is subject to the relevant University rules, regulations and policies.

Policy section 12.

13.1 Grounds required for any report or inquiry

• Detector output is never enough. Output from an AI detection tool, alone or combined with impressions of style, is not reasonable grounds to report or investigate anyone, and never alone the basis of a sanction
• Every report states at least one specific ground that can be checked against the work, its sources or the person's records
• Grounds at every AI use level: references that do not exist or do not contain the cited content; fabricated or falsified data, results, quotations or case details; figures that are internally inconsistent or irreproducible; invented facts typical of AI output (non-existent laws, cases, institutions or persons); evidence that Confidential or Restricted material was placed in an unapproved tool
• Evidence of AI use counts only where the level did not permit the use or it was not disclosed: leftover AI interface text; content matching AI output in the person's own records or identical to another person's work; an admission or records showing use beyond the level; or, for assessed work, a documented difference from the student's own supervised work (which justifies a meeting, and referral only if the student cannot explain or reproduce the work)
• Not grounds, alone or together: polished language, particular words or sentence patterns, uniform structure, an innocently explained change in writing quality, or writing by a non-native English speaker
• Only approved detection tools may be used, only by the person assessing the work, and results are Confidential. A detector score is a probability estimate with false positives and negatives; decisions rest on what can be checked

13.2 Inquiry into assessed work

Any lecturer may invite a student to a short conversation about how work was produced; this is not an allegation and creates no misconduct record. A lecturer who suspects misuse records the section 13.1 grounds and meets the student, who may explain the work and produce drafts, notes or AI records. If the explanation satisfies the lecturer, the inquiry ends with a brief file note; otherwise the case goes to the Head of Department under the examination and student disciplinary procedures. Refusing without good reason to attend or discuss the work is itself a ground for referral.

13.3 Screening of reports on staff and research

Reports on staff or research are first screened in confidence within ten working days, by the Director of Research, Innovation and International Collaboration (or a designate without conflict of interest) for research, or by the Registrar (or designate) for other staff matters. The screener dismisses, with written reasons, reports lacking section 13.1 grounds, reports on published work excluded by section 10.5, and reports judged against rules that did not apply when the work was produced. The person reported is not named or required to respond before a report passes screening.

13.4 Proportionate sanctions

13.5 Reporting, fair hearing and appeals

Anyone may report a suspected breach, or harm caused by an AI system, to a Head of Department, the Data Protection Officer, the AI Governance Committee secretariat or the whistleblowing channel; good-faith reporters are protected from victimisation. The identity of the person reported stays confidential during screening and inquiry. A report made maliciously, or deliberately based only on detector output about another person's work, is itself misconduct. No penalty is imposed until the person has been told the allegation and evidence, had a fair opportunity to respond and been told of the right of appeal.

Policy section 13.

• Ownership of works, inventions, software, models and datasets created with AI assistance is governed by the University's intellectual property policy, which will be reviewed to address AI-assisted works
• Because the status of purely AI-generated content under the Copyright Act 2022 is unsettled, human creative contribution is the basis of any claim to authorship or inventorship, and significant reliance on AI is disclosed in intellectual property and patent declarations
• Material placed in AI tools is used within its copyright and licence terms, including licensed library resources. Ownership questions go to the Legal unit

Policy section 14.

• Due diligence: before signing any agreement involving AI tools, research, data sharing or services, the University examines the partner's record on data protection, security, fairness and ethics, and settles roles, data ownership, intellectual property, publication rights, compliance with Nigerian law and exit arrangements, leaving academic decisions in its own hands
• Sponsored and donated tools: assessed for usability, accessibility, bias and fit with educational goals before deployment, with users told who sponsors them
• Transnational programmes: students face AI rules and integrity standards of comparable rigour at both institutions, partner tools comply with Nigerian law, access is equal, and AI quality assurance forms part of the agreement submitted for NUC approval
• Community initiatives: AI projects benefit the communities involved, respect local cultures and languages, and provide a way to give feedback

Policy section 15.

• Work towards device access, reliable connectivity and low-bandwidth options so AI does not widen gaps between students, and encourage AI tools and research serving Nigerian languages and cultural contexts
• Arrange assistive AI use for students and staff with disabilities through disability support services and record it, so it is not mistaken for misconduct
• Consult affected staff and provide retraining before any deployment that would substantially change roles, workloads or staffing
• AI tools may add to, but never replace, counselling, chaplaincy, health and pastoral services, and direct users to human help on health, safety or distress
• As stewards of creation, members of the University avoid needless AI use, since training and running AI systems consumes energy and water

Policy section 16.

• Continuing, funded AI training for academic staff, administrative staff and leadership, and AI literacy and integrity orientation for all new students, with minimum requirements set in the Implementation Framework
• Demonstrated expertise in AI teaching, research and innovation counts in staff appraisal and promotion, subject to the appointments and promotions bodies
• Planning and budgeting for connectivity, licensed tools, computing and cybersecurity, a dedicated AI budget line, and support for interdisciplinary AI research, innovation and technology transfer

Policy section 17.

The Vice-Chancellor constitutes a University AI Governance Committee, as the NUC Framework requires; it reports to Senate through the Vice-Chancellor. The Babcock University Health Research Ethics Committee (BUHREC) and other research ethics committees review AI use in research involving human participants or personal data.

Abridged: the policy also assigns duties to School and College AI Leads, the Postgraduate School, ethics committees, the University Library, the Research office, the Chaplaincy, student representatives and vendors. Policy section 18.

• The AI Governance Committee reviews the policy at intervals of not more than 12 months, and earlier if a new law, regulation or NUC directive is issued, AI technology changes significantly or a serious incident exposes a gap, taking account of students' and staff's views
• Changes to academic provisions require Senate approval; other changes require Management approval and Governing Council ratification. The Committee may update the Implementation Framework, registers and forms in line with the policy
• The policy is read with the University's research ethics, examination, academic integrity, intellectual property, ICT acceptable use, data protection and disciplinary instruments. On AI-specific matters this policy governs until those instruments are harmonised

Policy section 19.

• The policy commences on the date of ratification by the Governing Council. Assessments set before commencement continue under the rules announced when they were set
• Disclosure requirements apply to work submitted, published or funded after commencement; earlier work is judged by the rules and norms in force when it was produced. Fabrication, falsification and plagiarism remain misconduct whenever they occurred

Approval path: recommended by the drafting committee or AI Governance Committee → considered by University Management → approved by Senate → ratified by the Governing Council.

Policy section 20.

Nigerian legislation and regulations

• Nigeria Data Protection Act 2023, and the GAID 2025 (Nigeria Data Protection Commission)
• Copyright Act 2022
• Cybercrimes (Prohibition, Prevention, etc.) Act 2015, as amended in 2024
• National Code of Health Research Ethics

National frameworks and guidelines

• NUC, AI Framework Policy for Nigerian Universities (2025)
• NUC, Towards Responsible AI Integration in the Nigerian University System (2025)
• NUC, Guidelines on Transnational Education in Nigeria
• National Artificial Intelligence Strategy 2025–2029

International instruments and guidance

• UNESCO Recommendation on the Ethics of AI (2021); Guidance for Generative AI in Education and Research (2023); AI Competency Frameworks for Students and Teachers (2024)
• COPE position statement on authorship and AI tools (2023); CLUE recommendations on cooperation between universities and editors (2021)
• Perkins, Furze, Roe and MacVaugh, The Artificial Intelligence Assessment Scale (2024); APA Style guidance on citing generative AI (2025)

Babcock University instruments

The Implementation Framework for the Policy on AI, and the University's research ethics, examination, academic integrity, intellectual property, ICT acceptable use, data protection and disciplinary instruments.

Part 1. Course AI statement (completed by the lecturer for every course outline)

Lists the course code and title; the AI use level (0 to 4) of each assessment; uses allowed and not allowed in the course; the disclosure required at each level; approved tools suited to the course; and notes that breaches are dealt with under section 13. Standard wording is provided for each level, with the general note: where no level is stated, Level 2 applies to coursework and Level 0 to invigilated assessments; ask before using AI if unsure; keep drafts and AI records until the result is final and any appeal period has ended.

Part 2. AI disclosure form

Submitted by students with work at Level 3 or 4, or at Level 2 where the lecturer requires it. It records matriculation number, department and school, course, title of work, AI use level and date, then a table of each tool and version, date(s) used, purpose, parts of the work affected and how the output was checked or changed. The student declares that:

• Their use of AI stayed within what was permitted, and the table lists all of it
• They checked every fact, reference and quotation, and every cited reference exists and supports their statement
• The ideas, judgements and conclusions are their own
• They did not place others' personal data or confidential material in an AI tool
• They will produce prompts, outputs or drafts on request and may be asked to explain the work orally

Final-year projects, long essays and dissertations also carry a signed declaration of compliance, naming the AI tools used (or stating none was), countersigned by the supervisor.

Part 3. Citing AI in APA style

• A specific conversation readers can open through a share link is cited as a generative AI chat, with the AI company as author, the date, the chat title and the shared URL
• A tool in general, or a conversation that cannot be shared, is cited with the AI company, year, tool name, a description such as "Large language model" and the tool's URL
• Say in the text how the tool was used, usually in the methods section; prompts do not appear in the reference but should be kept in your records
• Never cite a source an AI tool suggests until you have found and read it yourself. Where a course, journal or funder prescribes another style, use it

Abridged; the full forms and wording are in Appendix A of the approved policy.

Adapted from the AI Compliance Checklist for Postgraduate Research in the NUC Committee Report. The researcher records name, matriculation number, programme (PGD, Masters, MPhil, PhD), department and school, research title, supervisor(s) and stage (proposal or final submission).

• Section A, declaration: whether AI tools or platforms were used in the research, or none were
• Section B, details of AI use: each tool and version, its purpose, the chapter or section affected, how outputs were verified, and a brief description of use (not more than five lines)
• Section C, compliance with ethical principles: transparency (use stated in the methodology or compliance statement); academic integrity (all interpretations and conclusions are the researcher's own); accuracy (facts, references, quotations and analytic outputs verified against primary sources); data privacy (no personal or sensitive participant data entered unless the tool was approved and covered by consent and ethical approval); bias and fairness (possible biases assessed and addressed); and informed consent (participants told when AI interacted with them or processed their data)

The checklist is signed by the student and countersigned by the supervisor.

Abridged; the full checklist is in Appendix B of the approved policy.