6.1–6.2 - Authenticity, attribution and responsible assistance

Independent achievement and authentication

Assessed project work must demonstrate your own independent achievement, and your centre needs evidence to authenticate it.

Keep meaningful intermediate work: early analysis, design decisions, versions of code, real test results and reasons for changes. Regular direct supervision allows the teacher to see how the work develops. Be able to explain what your program does, why you chose its structures and how a change followed from evidence. A finished file appearing suddenly is harder to authenticate than a natural progression the teacher has observed.

The student signs the Candidate Record Form to confirm the work is their own. Every teacher who marked it signs the authentication declaration, and the centre attaches the form to the work. The current AI guidance also requires appropriate reference to any AI used. Signing is a declaration with consequences, not a substitute for evidence or permission to submit someone else's output.

Only original independent achievement earns credit. A correctly functioning section obtained elsewhere does not become your own achievement merely because you can run it. External sources and tools may play legitimate roles, but those roles must be transparent and the submitted work must demonstrate your own decisions, understanding and development.

Your teacher retains professional judgement about authenticity and assessment. NovaLearn practice scores do not authenticate a project, and a tutor conversation is not an official assessment record for the complete NEA.

Guidance and assessed-work boundaries

The rules distinguish explaining the task and criteria from giving improvements to a student's assessed draft.

A teacher may clarify what the task requires and how its marking criteria work, and advise whether the proposed problem is suitable. When checking drafts, the specification says the teacher must not comment or suggest how the work could be improved. They may ask about the approach and highlight the criteria. These restrictions concern assessed project work; separate formative exercises can teach the underlying skills without producing the learner's submission.

Acceptable additional assistance must be recorded on the Candidate Record Form, and the awarded mark must represent the student's unaided achievement. The teacher still signs the required authentication statement; an unsigned statement means AQA cannot accept the work for assessment. Further guidance comes from the current subject guidance and teacher standardisation. Discuss uncertain assistance with the supervising teacher before relying on it.

Once work has been submitted for marking and marked, it cannot be returned to the student for improvement, even if no feedback or awarded mark has been disclosed. Keep this boundary clear when scheduling a final submission.

Use NovaLearn to learn general skills through different scenarios: explain why an objective is measurable, compare test choices or understand a data structure. Do not ask it to produce, refine or certify assessed project elements in a way that replaces evidence of your own skills. Acknowledging a source does not automatically make every form of assistance acceptable.

Attribution and responsible source use

A reader should be able to distinguish your own work from material and assistance obtained elsewhere.

Acknowledge copied material from books, websites, libraries, examples and other sources. Identify the source and make clear what was used, where it was used and what you changed or supplied yourself. Put quotation or code boundaries where appropriate instead of blending external text into your own prose. Keep enough reference information for the material to be located. A generic bibliography alone may not identify which code or decision depends on which source.

Do not submit work that is not your own, lend your work to other students, or allow others to access or use your independently sourced source material for their work. The specification also forbids unacknowledged copying and third-party word processing, and inappropriate, offensive or obscene material. These are malpractice concerns; possible penalties include disqualification. Follow the centre's rules when collaborating in ordinary learning so that assessed work remains independent.

An external library can be a legitimate dependency, but explain its role and your own contribution. If a package performs the main operation, do not describe that operation as entirely your implementation. Attribution makes origin transparent; the technical-solution criteria still judge your independent achievement rather than the library's complexity.

Keep a record of assistance as well as published sources. The teacher records work that is not the student's own on the Candidate Record Form or another appropriate location. Ask the teacher how to reference uncertain material before submission; do not assume that changing variable names or paraphrasing a passage removes its external origin.

Current AI guidance and transparent use

AQA's July 2026 guidance allows some legitimate NEA support uses of AI while requiring independent development, understanding, reflection and complete acknowledgement.

Examples include supporting initial research, generating questions for gathering requirements, producing interface mock-ups and generating synthetic test data. These are possible preparatory uses under centre policies, not permission to submit an AI-written project as your own. Direct AI content itself receives no credit. The learner must extend, adapt and develop material, demonstrate understanding through analysis and practical testing, and reflect critically on its quality.

For instance, AI-generated synthetic customer records may contain duplicate identifiers or invalid dates. Check the data against your planned cases, correct or select it deliberately, explain the changes and test the behaviour using it. Merely accepting every generated row provides weak evidence of decision making. AI explanations or code may be wrong or unsuitable; verify them through appropriate reliable sources and real tests rather than treating fluent output as authority.

Record the AI tool, version and access link, prompts or search terms, settings/parameters, annotated screenshots of its use and outputs, and a written account of how the content was adapted, refined and integrated. If the tool does not supply version/details, independently verify the generated content and reference the sources used for verification. The teacher supports embedding references in the work and on the Candidate Record Form where applicable. Poor or incomplete acknowledgement is a problem too.

Protect intellectual property and personal information. Never upload private information about yourself or others. Use lawful, suitable synthetic examples for practice, follow centre policies and do not allow learner work to train models without express permission. A tool's availability does not establish that its terms or privacy handling are suitable.

Unacceptable use includes presenting unmodified or unacknowledged generated code as your own, using AI to complete or refine work without evidence of your skills, copying or paraphrasing whole generated responses, and failing to reference its use. Even referenced, heavily derived work with minimal personal development lacks independent achievement. A reference is necessary when AI is used, but is not a guarantee that the use or final work is acceptable.

These rules concern NEA. They do not permit AI in the Paper 1 examination: the verified 2027 Teacher Notes update prohibits AI features in the examination IDE. Keep separate formative learning, permitted centre-approved project support and the examination conditions.

Concerns and professional judgement

Concerns about authenticity require the centre's procedures and evidence, rather than an automated verdict about who wrote a file.

Teachers may retain intermediate work and ask the learner to explain the progression between versions. A short discussion can explore whether the learner understands their decisions and changes. Unexpected work unlike the learner's usual output may need investigation, but it is not by itself proof of malpractice. AI detection output must not replace appropriate evidence or judgement.

If malpractice is identified before the student has signed the authentication declaration, the centre deals with it seriously under its internal procedures; it does not need to report that incident to AQA. After the declaration has been signed, the head of centre must report the case to AQA at the earliest opportunity using the required procedure, including JCQ/M1 where applicable. The current AI guidance also requires reporting detected or suspected AI misuse after signing. Consult the exams officer about the applicable process and do not conceal concerns by quietly replacing the submitted work.

Teachers must not accept work that is not the learner's own, and the head of centre is responsible for preventing inauthentic submissions. AI tools must not be the sole method of marking work. Professional judgement informed by observed skills and understanding remains essential, and marks must reflect the learner's independent achievement where content is heavily derived from AI.

For your own learning, preserve honest progress, explain sources and raise uncertainty early with the teacher. Neither a generated authenticity statement nor a formative practice score can replace the centre's supervision and signed declarations.

The primary guidance is AQA's AI guidance for the Computer Science NEA, version 1.0, July 2026, read alongside the specification and current centre guidance.