Tell people when they're talking to AI

• Your app has a chatbot, AI assistant or AI agent that talks to users.
• Your app generates images, audio, video or text that users publish, download or share.
• People in the EU use your app, or you operate in US states with bot or AI disclosure rules.

• From 2 August 2026, the EU AI Act (Art. 50) requires that people are told when they're interacting with an AI system, unless it's obvious, and that providers of generative AI mark synthetic audio, images, video and text in a machine-readable way. Deployers must disclose deepfakes and AI-generated text published to inform the public on matters of public interest, unless it was human-reviewed under editorial responsibility.
• California's bot disclosure law covers bots used to sell goods or influence votes, and Utah requires disclosure of generative AI in certain consumer interactions.
• Beyond the law, people trust products more when they know what they're talking to.

• A list of every place AI talks to users or produces content.
• Which AI provider you use, and whether user content is used for training under your plan.

Copy it into your AI builder. Fill in anything in [brackets] first, and review every change before you ship it.

Add clear, consistent AI disclosure to this app.

1. CONVERSATIONS
- Wherever a user chats with or gets replies from an AI model, show a visible line before or at the start of the conversation: "You're chatting with an AI assistant. It can make mistakes." Keep a small "AI" label on the assistant's messages.
- Don't give the assistant a human name, photo or wording that suggests it's a person. If a human can take over, say clearly when a human joins.
- If someone asks "Am I talking to a bot?", the assistant must say it is AI. Add this to the system prompt.

2. GENERATED CONTENT
- Label AI-generated images, audio, video and text in the interface ("Generated with AI").
- For downloads: add machine-readable marking where the format and your provider support it (for example C2PA content credentials or metadata for images, and a note in the file for text). Keep any marking your AI provider already adds; don't strip metadata.
- If users can publish AI-generated content publicly through your app, add a label they can't remove by accident.

3. EXPLAIN IT
- Add an "About AI in [app name]" section to the privacy notice or help: which features use AI, which provider, what data is sent, whether it's used for training, and how to contact a human.

4. SENSITIVE USES
- If the AI makes or supports decisions about people (hiring, lending, insurance, housing, education), stop and list these for me: they carry extra rules and need human review.

When you're done: list every file you changed, explain each change in one line, and give me a step-by-step test for each numbered item above. Don't change anything unrelated.

• A disclosure buried in the terms instead of shown in the chat.
• Human names and stock photos for AI assistants.
• Stripping metadata from generated images when resizing or compressing them.

1. 1Open the chat as a brand-new user: the AI notice is visible before the first message.
2. 2Ask the assistant 'Are you a human?': it says it's AI.
3. 3Generate and download an image: it's labelled in the app, and its metadata (for example via an online metadata viewer) still shows any provenance data.
4. 4The privacy notice or help page names the AI provider and explains training.

• EU AI Act (Regulation (EU) 2024/1689)

General guidance for common problems, not legal advice. Whether a rule applies depends on where you operate, who your users are and the size of your business.