App AI provenance — UX audit rules

Rule handbook metadata

Rule: DET.APP.AI_PROVENANCE · lane deterministic
Page status: current
page_version: 460a8845024f7b08b7379e2bb38e7b046b8a674a3f4a4192ed0574e384922393
generated_at: 2026-08-13T02:54:59.996Z
registry_fingerprint: aef1de6082cf0f50d463783c843dee0ffb9132fbd5ed4ea6e5bb3f031f359c72

How this rule is fixed

Not in the deterministic fixer pilot registry (stub or not yet wired). Use the After example below as the target state and remediate manually or extend pilot-registry.json.

Detection module: docs/design/ux-audit/deterministic-design-rules.md#det-app-ai-provenance. Scroll down for Before / After examples and Evidence and remediation steps.

Purpose

AI-assisted operator suggestions must disclose provenance and confidence and support human accept/reject/revert (ForgeAISuggestionReview).

Pass criteria

When [data-ai-generated], [data-ks-hash="Fai"], or .forge-ai-suggestion-review is present, the surface includes:

AI label (Fal / .forge-ai-label / data-ai-label)
Provenance panel (Fpv / data-ai-provenance)
Confidence indicator (Fci / data-ai-confidence)
Accept/reject and/or revert (Fra / matching button labels)

Evidence and remediation

Run tools/website-ux-auditor/auditor-tests/invoke-det-ruleset-harness.sh --only-rule DET.APP.AI_PROVENANCE. Implementation: tools/website-ux-auditor/design-rules/deterministic/generated/det-app-ai-provenance.check.js. Prefer createAISuggestionReview.