Safety & Design
AI disclosure
Clearly telling someone they are talking to an AI, or looking at AI-generated content, rather than a human or an unedited recording.
- Policymakers and staff
- Businesses
- Parents
What parents should know
AI disclosure means the product says it is a machine. Kids should not have to guess whether a kind reply is a person. The same idea covers images and audio that were generated. A label does not make the content true. It stops one kind of trick.
On this page
What is AI disclosure?
Disclosure is a notice at the point of use. In a chat, it means the interface and the bot's own words do not pretend to be a human friend. In media, it means a viewer can tell a clip was generated or heavily edited. The EU AI Act's transparency duties, in application from August 2, 2026, include telling people they are interacting with a chatbot.
The House-passed KIDS Act would also require AI chatbot providers to tell minor users that the chatbot is an AI system and not a human, and to share crisis-line information. That bill is not law. A product can disclose anyway. A disclosure buried in a terms-of-service link is weaker than a name and a role on the screen.
Why AI disclosure matters
Companion products blur the line on purpose. A child who thinks the bot is a kid at another school will share more and believe more. Saying assistant, every time, is a small disclosure with a large effect.
Fake audio and images need a second kind of label. Watermarks and content credentials are attempts. Until they are reliable, the household rule is still: a shocking clip is not proof, and a bot is not a classmate.
How it shows up in practice
- The chat header says the reply is from an assistant, and the bot does not claim to be a student at a nearby school.
- A generated image in a project is described as generated when it is shared.
- A voice on the phone asks for money, and the family treats it as unproven until they call the person back on a known number.
- A policy draft cites the EU transparency date and does not pretend the US bill has passed.
How HeyOtto helps
Otto is introduced as the assistant. HeyOtto does not present Otto as a human or a classmate. HeyOtto does not claim a formal EU AI Act disclosure filing. The product already says what Otto is. If a child is in distress, the chat can show crisis resources, including the idea of a real helpline, and a parent can be alerted. HeyOtto does not auto-contact emergency services.
- Tell children Otto is an assistant. The chats are readable, and Otto does not ask them to keep secrets.
- Parents can read the chat and correct the story if a child treats it like a friend.
- The KIDS Act disclosure duty is a bill. Today's wording is a product choice.
For families
Try freeFAQs
Does the EU already require chatbot disclosure?
The European Commission says the AI Act's transparency rules apply from August 2, 2026, including for people interacting with chatbots. High-risk duties are on a later date. Whether a given US product is in scope is a legal question. This page is general information, not legal advice.
Is a disclosure the same as saying the answer is true?
No. A label says who or what produced the content. It does not check the facts. You still open a source, and you still treat a generated image as a drawing.
What should I say at home?
Otto is the assistant. It is not a friend and it is not another child. You can read the chats. If something it says scares you, we look at it together and we can call a real person.
Are watermarks a form of disclosure?
They are one technical attempt to mark AI-made media. They fail when the file is cropped or re-encoded, and many images have no mark. A visible label plus a household habit of checking beats a hidden mark you cannot see.
Sources
Last reviewed September 26, 2026. This entry is reviewed twice a year.