If the AI wrote it, whose fault is the mistake?
AI wrote it, but your name is on it.
The lawyers in Mata v. Avianca were sanctioned. ChatGPT wasn't (see "AI's sources need checking too").
Deloitte refunded part of its fee for the report with made-up references. The model didn't (see "The nicer AI's work looks, the less we check it").
When Air Canada's website chatbot gave a customer wrong information about bereavement fares, a tribunal ruled that Air Canada had to honour it. The airline had argued the chatbot was responsible for its own words. That argument didn't work.
Every AI literacy framework we looked at, from Anthropic's to the UK and US governments' and the EU's, says the same thing in different words: the person who uses the output is responsible for it.
Anthropic's course calls this Diligence: being honest about how you used AI and taking responsibility for what you share.
In the EU, the AI Act requires organisations that use AI to make sure their staff have enough AI literacy. That duty has applied since 2 February 2025, and national regulators can enforce it from 2 August 2026.
The European Commission's own guidance gives an example: staff using ChatGPT should be told about specific risks such as hallucination.
Being open about AI use is part of owning it. Whether you need to say so depends on the situation: your company's policy, your profession's rules, or simply what the reader would expect.
A useful test: if the reader would feel misled when they found out, tell them up front.
Stop and get a person involved when:
Read it as if your name is on it, because it is.
Disclose when the reader would expect you to.
Keep a human in the loop for decisions about people.