Disclosing AI Use at School and at Work

When should you tell people you used AI? In 2026, this is one of the most practically important questions a student faces. Get it right and you stay safe in school, build trust at work, and develop a reputation as a responsible practitioner. Get it wrong and you risk academic misconduct, broken trust, or legal exposure.

What You'll Learn

• The current state of AI disclosure norms in academia and the workplace
• Three rules of thumb that work in 90% of situations
• How to write disclosures that are clear, fair, and protective
• A reusable template you can use for school and work

Why Disclosure Matters

AI tools are now so good that you can produce work that appears to be your own with minimal effort. This creates three intertwined problems:

1. Honesty. Pretending AI work is your own is a form of misrepresentation.
2. Quality assurance. AI hallucinates; readers benefit from knowing where to verify.
3. Accountability. If an AI-produced section turns out to be wrong, the chain of responsibility should be clear.

Disclosure addresses all three.

The Three Rules of Thumb

When in doubt, fall back to these:

1. If a teacher or employer asked, would I be comfortable saying yes? If not, you should disclose proactively.
2. Would this matter if discovered later? If discovery would cause embarrassment, accusation, or legal exposure, disclose now.
3. Could a reasonable person object to my use? If yes, disclose.

These three together cover almost every real situation.

Disclosure in Academic Settings

Universities have moved fast to set policies in 2024–2026. The current spectrum looks like this:

Stance	What it usually means
Prohibited	No AI use on this assignment, period
Restricted	AI okay for some tasks (research, brainstorming) but not the final writing
Allowed with disclosure	Use freely, document how
Encouraged	AI literacy is part of the assessment

Your professor's syllabus will state the policy. If it doesn't, ask — by email, in writing — before you start. "Is AI assistance allowed on this assignment, and if so what kind?" That single email protects you better than any defense after the fact.

If AI is allowed with disclosure, a strong disclosure looks like:

AI assistance disclosure: "I used Claude Opus 4.7 to brainstorm an outline and to suggest counterarguments for Section 3. I wrote all final prose myself and verified all cited sources independently. Total AI-assisted time: roughly 30 minutes."

Specific, honest, and shows you used AI thoughtfully.

Disclosure at Work

Work is messier. Most companies are still writing their AI policies. Some general patterns:

• Internal drafts and brainstorming: Often fine without disclosure unless the company has a strict policy.
• Customer-facing content: Increasingly expected to be disclosed (especially in regulated industries).
• Code: Many engineering teams now require commit messages to mention AI assistance.
• Decisions about people (hiring, performance, customer credit): Disclosure may be legally required (EU AI Act, NYC bias audit law, several U.S. state laws).
• Marketing claims: Some jurisdictions now require disclosure of AI-generated images of people.

When you join a team, ask: "Is there an AI use policy? Are there situations where I'm expected to disclose AI assistance?"

When Lack of Disclosure Is a Bigger Risk Than the AI Itself

Several real cases have made the news in 2024–2025 where disclosure (or lack of it) was the actual problem:

• A consulting firm submitted a government report that included AI-fabricated citations. The firm's reputational damage came not from using AI but from not disclosing it and not catching the fabrications.
• A tech company released a "personalized recommendation" feature that was actually using a third-party generative AI. When this came out, customers felt deceived even though the feature worked fine.
• A novelist won a major literary prize and was later revealed to have used AI throughout. The community reaction was harsher than it would have been with upfront disclosure.

The pattern: people often forgive AI use, but they rarely forgive concealment.

What Disclosure Should Contain

A good disclosure has four elements:

1. Tool used. "Claude Opus 4.7" or "ChatGPT 5" — name and version.
2. What it was used for. Be specific: "drafting," "summarizing sources," "translating," "code review."
3. What human review was done. "I reviewed and edited all output and verified citations."
4. Limitations or caveats. Optional but appreciated: "Some sections rely on AI summarization; check sources for full nuance."

Avoid vague language like "AI was used to assist with this work." It tells readers nothing.

Hands-on: Write Three Disclosures

Imagine you used Claude to help with each of the following. Write a one- or two-sentence disclosure for each:

1. A research paper for a college course (the policy says "AI allowed for brainstorming and outline only")
2. A LinkedIn post about your internship project
3. An internal Slack message proposing a new feature at your job

Then post your disclosures into a chatbot and ask:

"Critique these three AI-use disclosures. Are they specific enough? Honest? Professional? Suggest improvements."

Save the strongest version as a template.

A Template You Can Reuse

AI Use Note
Tool: [Name + version + date]
Purpose: [One sentence — drafting, summarizing, brainstorming, code, translation]
Human review: [What you did to validate output]
Notes: [Any limitations the reader should know about]

Drop this at the bottom of academic papers, project reports, longer LinkedIn posts, or commit messages. Over time, you will become someone whose AI use is trusted because it is documented.

When AI Disclosure Is Mandatory by Law

A short list of cases where you may be legally required to disclose, depending on your jurisdiction:

• Any deepfake or AI-generated image of a real person used in advertising (several U.S. states, EU AI Act)
• AI-generated political ads (U.S. FEC, multiple states, EU)
• AI use in hiring decisions (NYC Local Law 144, EU AI Act high-risk systems, Illinois AI Video Interview Act)
• AI use in medical, financial, or legal services (various sector-specific rules)
• AI-generated content that interacts with EU consumers (EU AI Act transparency obligations)

This list is growing. Treat any high-stakes use of AI as potentially regulated.

Key Takeaways

• Three rules of thumb cover most disclosure decisions.
• For school: read the syllabus, ask in writing, save the response.
• For work: know your company's AI policy and ask if it doesn't exist yet.
• A strong disclosure names the tool, the use, the human review, and the caveats.
• Concealment hurts careers more than the AI itself ever would.