UCL's 2026-27 GenAI Rules: What Changed in the Academic Manual

Four things, all in the student academic misconduct procedure. We compared the two combined Academic Manuals in full, the 2025-26 edition (698 pages) and the 2026-27 edition (731 pages), on 25 September 2026:

UCL's own page of changes for 2026-27 summarises this chapter in one line: "Section 9: Student Academic Misconduct Procedures replaced (EdCom May 2026)". It does not mention GenAI, the viva, the removed promise or the relocated duty.

The 2025-26 manual listed three things that "will not be considered Academic Misconduct". The third was this: "Use of GenAI tools that does not exceed that permitted in the assessment brief: each module leader is responsible for informing students of the extent to which they are permitted to use GenAI tools in each assessment. Evidence of use where there has not been a clear direction from the module leader will not be penalised."

That item did two jobs, and the 2026-27 manual treats them differently. The duty survives and moves to the start of the procedure: "module leaders are responsible for providing accurate and up-to-date information to students about their assessments. This includes information about appropriate referencing, clear instructions about collaborative working, the scope of permissible use of Generative Artificial Intelligence (GenAI)", with a link to the misconduct regulations. The promise does not survive. Nothing in the 2026-27 manual says what happens when that information was never given.

Read that carefully. It does not say that GenAI use under an unclear brief will now be penalised. It says the manual no longer promises that it will not be, while still making it the module leader's job to tell you.

The 2026-27 manual defines it as "Unauthorised and/or unacknowledged use of Generative Artificial Intelligence (Misuse of GenAI): defined as the use of assistive tools beyond the remit of the assessment brief, or to create false data, text, or references."

Three details are worth noticing. The definition speaks of "assistive tools", not only chatbots, so the test is what the brief allows, whatever the tool is called. Using GenAI to fabricate data sets or references was already misconduct in 2025-26, listed under Falsification; it now sits inside Misuse of GenAI, and "text" has been added. And the category name has two halves, unauthorised and unacknowledged: use the brief permits can still be a problem if you do not declare it the way the brief asks.

One exemption is word for word the same in both editions. "Language and Writing review" is not academic misconduct, "defined as having a third-party or software check areas of academic writing such as structure, fluency, presentation, grammar, spelling, punctuation, and language translation." It continues: "However, this may be considered Academic Misconduct if substantive changes to content have been made by the reviewer or software or at their recommendation, which would suggest that the reviewer had either produced or determined the substantive content of the submission, or, in cases of language translation, if the student is being assessed on their ability to translate or use a language other than English."

So a grammar and fluency check is inside the line, a tool that changes what your work says is not, and translation is not covered when the assessment is testing your ability to translate or to use a language other than English.

For minor cases, the 2026-27 procedure gives the department three routes at Stage 1: apply a penalty, refer to an Investigatory Viva where authorship is in question, or refer to a panel. A first offence involving Misuse of GenAI is listed among the examples of minor cases. Where a viva is held, its purpose "is to establish the student’s authorship of the submitted work and make a decision on the case when it is Minor." The mechanics are specific:

• You must receive the invitation, the misconduct report form and the evidence "at least five working days before the scheduled Investigatory Viva".
• "A student’s attendance at an Investigatory Viva is mandatory and cannot be conducted in their absence." A student who does not attend may be referred to a Mandatory Faculty Interview and given a new date, and "Continued non-engagement with the Viva process may result in Suspension or Termination of Studies via the Mandatory Faculty Interview process."
• It "should be organised within 20 working days from the date of the Stage 1 outcome, unless there are reasonable circumstances that prevent this."

Earlier in the process there is a new deadline too: if you do not respond to an allegation or engage in a meeting "within 10 working days from the date of notification of the allegation, the Department/Faculty will make a decision on how to proceed in absence of a response." Asked to explain your own paper is a preparation checklist for this kind of meeting.

1. Find the GenAI line for each assessment. The 2026-27 definition turns on "the remit of the assessment brief".
2. If there is none, ask the module leader and keep the reply. Telling you is their responsibility under the 2026-27 procedure; the reply is also your record.
3. Declare what you used, the way the brief asks. "Unacknowledged" is half of the category's name.
4. Keep your drafts and notes. Version history helps, but it is a list of saved states rather than a record of every keystroke; what Google Docs version history records explains what it does and does not hold.
5. Answer any allegation within 10 working days. After that, the department may decide without you.
6. Never let a tool supply references or data. That limb of the definition applies whatever the brief allows.

By the 2026-27 definition HumanPen is an assistive tool: it rewrites English sentences. On a UCL assessment it is only usable as far as that assessment's brief allows, and any use of it should be declared the way the brief asks.

What it controls is which passages get rewritten, the ones flagged in a Turnitin or iThenticate report that you confirm before anything runs. It makes no claim about how a detector or a panel will read the result.

Does the 2026-27 procedure apply to me if I started at UCL before this year? Section 9 says "These procedures apply to all current and former students of UCL", and the page is marked "Published for 2026-27". If an allegation concerns work you submitted in 2025-26, ask your department which edition it is being handled under.

Is Grammarly allowed? Spelling, grammar and fluency checks fall under the "Language and Writing review" exemption, unless the tool makes substantive changes to your content. Rewriting features that change what you say are a different matter, and the brief decides.

Does UCL use AI detection scores as evidence? The 2026-27 manual does not say. It mentions Turnitin once, in a list of documents sent to a panel ("the student’s submission, Turnitin Reports, etc"), and contains no clause on what weight an AI detection score carries. Words beginning "detect" appear four times in the whole manual, all about the UCAS Similarity Detection Service in admissions.

My module leader never mentioned GenAI. What now? The 2025-26 manual said use without clear direction would not be penalised. The 2026-27 manual no longer says what happens, but it does make telling you the module leader's responsibility. Ask before you submit, and keep the reply.

• UCL Academic Manual, Section 9: Student Academic Misconduct Procedure ("Published for 2026-27"), read 25 September 2026.
• UCL, Academic Manual 2026-27, combined edition (PDF, 731 pages), read 25 September 2026.
• UCL, Academic Manual 2025-26, combined edition (PDF, 698 pages), linked from UCL's 2025-26 archive page, read 25 September 2026.
• UCL, Changes to the Academic Manual for 2026-27, read 25 September 2026.