Transparency Notice Template (EU AI Act)

Legal basis: Art. 50 transparency obligations (including disclosure of AI interaction and certain synthetic/deepfake content), with implementation support from related transparency provisions in the Regulation.
Use: Chatbots, AI-generated content, synthetic media/deepfake contexts, emotion/biometric-related disclosures where applicable.

1) Notice Metadata

• Notice ID: [TN-YYYY-###]
• Version: [vX.Y]
• Effective date: [YYYY-MM-DD]
• Channel: [Web/App/Email/Voice/Kiosk]
• Language(s): [EN/DE/FR/...]
• Owner: [Role]

2) Core Disclosure Block (Mandatory)

"You are interacting with an AI system operated by [Company]."

"This output may be generated or assisted by AI."

"If you need a human review, contact [Support channel]."

3) Context-Specific Notices

3.1 Chatbot interaction

• Statement: "This assistant is automated and may make errors."
• Human handoff path: [Link/process]

3.2 AI-generated text/image/audio/video

• Statement: "This content is AI-generated or AI-edited."
• Provenance marker used: [Label/watermark/metadata]

3.3 Synthetic/deepfake-like media

• Statement: "This media has been synthetically generated or manipulated."
• Placement: [Pre-roll/overlay/caption]

3.4 Emotion recognition / biometric context (if applicable)

• Statement: "This service may infer signals from biometric or behavioral inputs where legally permitted."
• Legal basis and safeguards reference: [Policy link]

4) Accessibility & Delivery Requirements

• Notice appears before or at interaction start
• Contrast and readability checks passed
• Screen-reader compatible version available
• Mobile and desktop placement tested
• Local language adaptation completed

5) User Rights and Contact

• Human review request: [Channel]
• Complaint/escalation channel: [Email/form]
• Privacy contact: [DPO/contact]
• Additional info link: [Policy URL]

6) Operational Checklist

• [ ] Notice text approved by legal/compliance
• [ ] Product implementation verified in production
• [ ] A/B variants documented (if any)
• [ ] Monitoring in place for broken notice rendering
• [ ] Review date scheduled

7) Common Mistakes

1. Hiding notices in terms only
2. Using ambiguous wording ("smart assistant" instead of "AI")
3. No human handoff option in high-impact contexts
4. No localization for EU users

8) Fillable Ready-to-Use Snippets

• Chat UI banner: "You are chatting with an AI assistant from [Company]."
• Email footer: "Parts of this message were generated with AI assistance."
• Media label: "AI-generated content"