Generator
AI Disclosure Generator
Several jurisdictions now require you to say plainly that a visitor is talking to software, and more are drafting. This writes the wording, says where it goes, and notes what each regime actually asks for.
- Wording and placementBoth, not just the text
- Regime awareEU AI Act, California, Colorado
- Not legal adviceSaid plainly, up front
Work it out
Your numbers
Your system prompt
This is a drafting aid, not legal advice. Regulation in this area is moving quickly; have your own counsel confirm before it goes live.
What the numbers mean
Disclosure belongs at first contact, not in a policy
The common requirement across regimes is that a person knows before they start, not that the information exists somewhere. A link to a policy page rarely satisfies it.
A human name and a stock photo is the risk
Most disclosure failures are not missing text. They are an interface designed to imply a person: a first name, an avatar, a typing indicator. Those undo the sentence.
Answer the robot question honestly
If a visitor asks whether they are talking to a person, the assistant must say no. This is the one behaviour regulators consistently test.
Keep a record of the wording
If the disclosure is ever questioned, you need to show what was displayed and when. A dated screenshot per release costs nothing and settles the argument.
Questions
No. It is a drafting aid that summarises commonly cited requirements. Regulation here is moving fast and varies by jurisdiction; have counsel confirm.
Some regimes exempt cases where it is obvious, but obvious is judged from the visitor's side. An explicit sentence is cheap and removes the argument.
Pick the strictest regime you serve and comply with that. It is simpler than conditional wording and usually covers the others.
Want this run against your actual site?
Tell us the page your assistant lives on and the questions you care about. A person runs them and sends back the raw answers.