TOPIC 04 OF 06
Transparency and disclosure
Telling people it's AIA distinct, lighter-touch duty applies broadly: people generally have a right to know when they're interacting with an AI system or looking at AI-generated or manipulated content. This is separate from permission and separate from the heavier high-risk duties — it's about honesty. Chatbots should not pose as human, and synthetic media (deepfakes, generated images) increasingly must be marked as such.
A support chatbot opens by making clear it's an automated assistant, not a person. A marketing image generated by AI carries a label. Neither is burdensome, and skipping them erodes exactly the trust the tool depends on.
Disclosure answers 'is this AI?' — it does not answer 'were we allowed to use what's in it?' Don't let a transparency label stand in for the copyright, likeness, or fairness questions it doesn't touch.
Check one AI-facing thing you run: is it obvious to the person that it's AI, and is generated content marked? If not, that's a cheap, high-trust fix.