Legal

How the AI Toolkit meets in-product AI transparency and disclosure requirements.

In placeLive in the toolkit today.PlannedComing in a near-term follow-up.Handled elsewhereOwned by platform or product teams, outside the design system.

Requirement coverage

Requirement	Covered by	Status
Show the user they are working with AI	Mark and wordmark	In place
Mark AI-generated content visibly	AI generated badge	In place
Prompt the user to review AI output	AiCaveat, on recommendation, action, and AI-content surfaces	In place
Show where the AI’s information came from	Source line in the activity timeline	In place
State what the AI can and cannot do	Scope line (AI Chat pattern)	Planned
Let the AI signal when it is unsure	Uncertainty statement (AI Chat pattern)	Planned
Warn before sensitive data goes to a tool	Data-clearance and egress warnings (AI Chat pattern)	Planned
Confirm before the AI takes an action	Action confirmation (AI Chat / actions pattern)	Planned
Collect feedback with a disclosure	Feedback vote widget disclosure	Planned
Data masking, access control, audit logs and retention, retrieval grounding, bias testing, human-in-the-loop configuration, approval workflows	Owned by platform, ML, and product teams	Handled elsewhere

Disclosure strings

String	Basis	Verbatim	Where
”The output is AI generated. Please review.”	Human-review flag	Yes, verbatim	AiCaveat default copy
”AI generated” (badge label)	Transparency marking	Yes, as the label	Badge
Source line, e.g. “From [sources]“	Explainability (a capability, no fixed phrase in the guidance)	N/A, our own UI label	Timeline marker

Source and legal basis

Based on UiPath’s internal in-product compliance guidance and the standards below.

Legal basis: EU AI Act (Regulation (EU) 2024/1689), EU GDPR (Regulation 2016/679), the EU Charter of Fundamental Rights, the OECD Principles on AI, the EU HLEG Trustworthy AI Guidelines, and ISO/IEC 42001:2023.

The requirement mapping and status reflect this team’s interpretation of that guidance.