Pattern library

Disclosure patterns for AI features

Curated UX patterns organised by category, with legal reference chips, status and pattern strength. Search and filter to find what fits your feature.

Legal references
Triggers
Showing 18 of 18 patterns

Entry-Point Notices

Up-front disclosures shown when a user first enters an AI experience, before expectations are set.

Chat window with a top banner: 'You're chatting with an AI assistant.'
Production ready
Strong
Entry-Point Notices

First Interaction AI Notice

Tell users at the start that they're talking to an AI.

Article 50(1)Article 50(5)
View Pattern

Inline & Contextual Notices

Small, local explanations placed next to AI controls or content so users understand what is happening in context.

Text field with an 'AI' info icon revealing an inline tooltip.
Production ready
Subtle
Inline & Contextual Notices

Contextual AI Tooltip

Explain what an AI element does, right where it appears.

Article 50(1)Article 50(5)
View Pattern
Chat composer with subtle helper text: 'AI can make mistakes.'
Production ready
Subtle
Inline & Contextual Notices

AI Limitations Notice

Set realistic expectations about accuracy and scope.

Article 13Article 50(5)
View Pattern

Persistent Labels

Always-visible badges and labels that keep AI involvement clear across repeated product use.

Button labelled 'Summarize' with a small pill badge 'AI' beside it.
Production ready
Medium
Persistent Labels

AI Feature Inline Badge

Persistent 'AI' label on features powered by AI.

Article 50(1)Article 50(5)
View Pattern
Footer with a 'How we use AI' link among other policy links.
Production ready
Subtle
Persistent Labels

AI Use Transparency Footer Link

Site-wide link to your AI usage and disclosure policy.

Article 50(5)Code of Practice
View Pattern

Output-Level Disclosures

Disclosures attached to generated or manipulated outputs, especially when content may travel outside the product.

Article card with a small 'AI generated' pill on top.
Production ready
Medium
Output-Level Disclosures

AI Generated Content Badge

Label content produced by generative AI.

Article 50(2)Code of Practice
View Pattern
Image preview with a corner watermark and info line 'AI watermark embedded'.
In review
Medium
Output-Level Disclosures

Synthetic Media Watermark Notice

User-facing notice that media carries a machine-readable marker.

Article 50(2)Code of Practice
View Pattern
Draft card with 'AI generated' label and timestamp above the body.
Production ready
Medium
Output-Level Disclosures

Output-Level AI Disclosure

Attach a disclosure to each AI-generated output, not just the feature.

Article 50(2)Article 50(5)
View Pattern

Explanation & Decision Support

Patterns that help people understand AI-assisted recommendations, limits, and decisions that may affect them.

Card with 'Why this?' link opening a small explanation popover.
Production ready
Medium
Explanation & Decision Support

Why This Recommendation Panel

Explain, in plain language, why the AI recommended something.

Article 13Article 86
View Pattern
Modal titled 'How this decision was reached' with inputs and appeal button.
Legal review recommended
Strong
Explanation & Decision Support

Decision Support Explanation Modal

Detailed explanation for consequential AI-supported decisions.

Article 13Article 86
View Pattern
Chat header with a 'Talk to a human' button next to AI notice.
Production ready
Medium
Explanation & Decision Support

Human Escalation CTA

Always offer a clear path to a human.

Article 50(1)Article 86
View Pattern

Public Interest & Editorial Content

Disclosure approaches for deepfakes, public-interest text, and editorial contexts where provenance matters.

News article header with an 'AI-generated' notice above the headline.
Legal review recommended
Strong
Public Interest & Editorial Content

Public Interest Text Disclosure

Disclose AI-generated text on matters of public interest.

Article 50(4)
View Pattern
Byline area showing 'AI generated · Human reviewed'.
Production ready
Medium
Public Interest & Editorial Content

Human Reviewed Label

Signal that a human editor reviewed the AI output before publication.

Article 50(4)
View Pattern
Footer line: 'Editorial responsibility for AI content: Publisher Name.'
In review
Medium
Public Interest & Editorial Content

Editorial Responsibility Notice

Clarify who is editorially responsible for AI-generated public content.

Article 50(4)
View Pattern

High-Risk / Legal Review Patterns

Stronger UX interventions for sensitive or consequential AI features where legal review is likely needed.

Video player with a prominent banner: 'Deepfake — synthetic media.'
Legal review recommended
Strong
High-Risk / Legal Review Patterns

Deepfake Disclosure Banner

Clear, prominent banner for deepfake media.

Article 50(4)Code of Practice
View Pattern
Video call screen with a top banner: 'Emotion analysis is active.'
Legal review recommended
Strong
High-Risk / Legal Review Patterns

Biometric / Emotion Exposure Notice

Tell people before they're analysed by emotion or biometric systems.

Article 50(3)Article 13
View Pattern
Callout: 'Legal review recommended' with a checklist beside it.
Legal review recommended
Strong
High-Risk / Legal Review Patterns

Legal Review Required Notice

Internal-flag pattern for features that need legal sign-off.

Article 13Article 50(3)Article 50(4)
View Pattern

Agentic AI Action Patterns

Confirmation and control patterns for AI agents that can take actions on a user's behalf.

Confirmation card showing pending agent action with Confirm / Cancel.
Production ready
Blocking
Agentic AI Action Patterns

AI Agent Action Confirmation

Confirm consequential agent actions before they run.

Article 50(1)Article 13
View Pattern