Universities are increasingly asking students and staff to be transparent about generative AI. Students may be required to state whether they used AI, explain how they used it or retain evidence of their process. Staff may also be expected to explain AI use in teaching, assessment, research or administrative work.<\/p>\n\n\n\n
In principle, disclosure is a sensible approach. It allows AI use to be judged in context rather than treating every interaction with a tool as the same. But disclosure depends on something that policy documents cannot create on their own. It depends on people feeling safe enough to tell the truth.<\/p>\n\n\n\n
If legitimate AI use is mocked in offices, meetings or corridor conversations, transparency can begin to feel less like responsible practice and more like personal risk. When that happens, disclosure starts to feel like confession.<\/p>\n\n\n\n
Generative AI deserves serious criticism. It can produce confident errors, fabricate sources, reproduce bias and be used to bypass learning. It raises difficult questions about privacy, copyright, environmental cost, labour and the meaning of independent work. Universities should not minimise any of these issues.<\/p>\n\n\n\n
The problem begins when criticism of the technology slides into judgement about the people who use it. There is an important difference between saying that a particular use is inappropriate and implying that anyone who uses AI is lazy, fake, less intelligent or less academically serious.<\/p>\n\n\n\n
Universities should be capable of maintaining that distinction. In computer science and related disciplines, we teach students to examine systems, assumptions, risks and consequences with precision. The same discipline should apply to our own conversations about generative AI. Slogans about the character of AI users do not help us assess what a tool is doing, whether the use is legitimate or what human judgement remains.<\/p>\n\n\n\n
A major weakness in polarised discussions about AI is that very different practices are collapsed into a single category.<\/p>\n\n\n\n
A student who submits AI-generated material they do not understand is not doing the same thing as a student who asks AI to generate revision questions. A colleague who sends an unchecked AI-generated document is not doing the same thing as a colleague who uses a tool to organise notes, simplify a piece of routine communication or test alternative wording before reviewing it carefully.<\/p>\n\n\n\n
The distinction matters for academic integrity, but it also matters for culture. If all AI use is treated as evidence of poor ability or weak professional standards, people who are using it within legitimate boundaries have an incentive to keep quiet. That makes it harder for institutions to understand actual practice and harder for colleagues to learn from one another about appropriate and inappropriate uses.<\/p>\n\n\n\n
This is especially important because disclosure is becoming a central part of institutional responses to generative AI. Students are often expected to state whether AI contributed to an assessment and what role it played. That approach assumes that students believe an honest answer will be interpreted fairly.<\/p>\n\n\n\n
The wider evidence suggests that this cannot be assumed. The 2025 HEPI and Kortext student generative AI survey reported that 92 per cent of surveyed students used AI in some form, compared with 66 per cent in 2024. Research on AI-use declarations has also reported substantial non-compliance, including a study in which 74 per cent of students failed to declare AI use despite a declaration being required. (Freeman, 2025). (Gonsalves, 2025).<\/p>\n\n\n\n
These figures do not tell us that shame is the only reason for non-disclosure. Students may misunderstand rules, forget requirements or deliberately conceal inappropriate use. But they do show that widespread AI use and effective disclosure are not the same thing. A declaration box is not a substitute for trust.<\/p>\n\n\n\n
Discussion of AI shame has often focused on students, but staff culture deserves equal attention. Academics are expected to model careful and transparent practice for students. That becomes difficult if staff themselves feel embarrassed to acknowledge legitimate use. (Giray, 2024).<\/p>\n\n\n\n
Generative AI can be used for focus, organisation, language, accessibility, routine drafting and communication without replacing the person’s ideas or judgement. For some people, these functions may be particularly helpful when workload is high, when they are working in an additional language or when they benefit from support with planning and organisation.<\/p>\n\n\n\n
That does not mean every use is appropriate. It means the value and risk of AI depend on what it is doing in the workflow. A healthy academic culture should make it possible to say, openly, that a tool was used for a limited purpose and then discuss whether that use was sensible.<\/p>\n\n\n\n
Shaming language can have uneven effects. Generative AI use may intersect with disability, neurodivergence, language background, confidence, caring responsibilities, workload and access to informal academic support.<\/p>\n\n\n\n
This does not justify misconduct and it does not mean AI should be recommended to everyone. It means institutions should avoid making tool use a proxy for intelligence or character. Someone who uses AI to organise their own ideas may be exercising judgement rather than avoiding it. Someone who uses it to create an accessible version of material may be addressing a barrier rather than seeking an unfair advantage.<\/p>\n\n\n\n
Policies need to be capable of recognising these distinctions. Culture needs to be capable of discussing them without ridicule.<\/p>\n\n\n\n
If universities want meaningful transparency, several conditions need to be in place.<\/p>\n\n\n\n
Students need assessment briefs that explain permitted and prohibited uses in concrete terms. Staff need confidence that they understand institutional expectations and can explain them consistently. Disciplines need examples that reflect their own forms of assessment rather than relying only on generic statements. Accessibility and equity need to be part of the policy design from the beginning.<\/p>\n\n\n\n
The language used around AI also matters. Responsible practice is more likely when people can ask questions, admit uncertainty and describe what they have done without expecting an immediate judgement about their competence. That does not weaken standards. It creates better information on which standards can be applied.<\/p>\n\n\n\n
Universities should also separate disclosure from automatic suspicion. A declaration should provide context for evaluating the work, not function as an admission of wrongdoing. If students or staff believe that saying they used AI will automatically make their work appear less legitimate, non-disclosure becomes a predictable response.<\/p>\n\n\n\n
The policy challenge around generative AI is not simply whether a tool is allowed or banned. It is whether universities can create environments in which its use can be discussed accurately.<\/p>\n\n\n\n
That requires clear boundaries and real accountability. It also requires restraint in how we talk about people. Critique should focus on the practice, the evidence, the risks and the remaining human responsibility. It should not divide academic life into real thinkers and AI users.<\/p>\n\n\n\n
Disclosure will not work if AI use feels like confession. If universities want students and staff to be transparent, they also need to build the conditions that make transparency safe. Generative AI deserves scrutiny. The people using it deserve clear guidance, education and respect.<\/p>\n\n\n\n
The sources below are those used in, or directly relevant to, the original article and the claims retained in this blog adaptation.<\/p>\n\n\n\n
Freeman, J. (2025). Student Generative AI Survey 2025. HEPI Policy Note 61, in partnership with Kortext. https:\/\/www.hepi.ac.uk\/reports\/student-generative-ai-survey-2025\/<\/a><\/p>\n\n\n\n Giray, L. (2024). AI Shaming. The Silent Stigma among Academic Writers and Researchers. Annals of Biomedical Engineering, 52, 2319\u20132324. https:\/\/doi.org\/10.1007\/s10439-024-03582-1<\/a><\/p>\n\n\n\n