Ad creative compliance gate

Business value

• Reduces the rate of ad disapprovals that stall campaigns. Platform reviewer flags on synthetic-content rules are slow to appeal; pre-flighting saves the days.
• Catches creatives that would require disclosure labels on platforms that mandate them — and adds the labels before submission, not after rejection.
• Documents AI use in the creative for compliance teams as the disclosure landscape evolves (FTC, state-level laws, platform rules).

Agent job to be done

Be a paid-media QA reviewer with platform-policy knowledge. Allow obviously authentic creatives. Hold creatives with AI-generation cues for disclosure-decision review. Reject creatives that would clearly fail platform synthetic-content rules.

format: social_postintended_use: publishdomain: ad creative compliance / paid acquisition

When to call VeracityAPI

After creative generation/upload, before campaign activation or platform submission. Re-run after any creative variant change.

What image URL to submit

Public HTTPS image URL for each creative variant. For carousel and dynamic creatives, score each frame independently.

Decision policy

• allow: low risk, no disclosure-trigger evidence, no platform-specific policy concerns.
• revise: medium risk — add an AI-disclosure label on platforms that require it, or replace with non-AI variant.
• human_review: high risk on regulated industries (financial services, healthcare, political advertising) where AI disclosure is mandatory and platform-policy review is rigorous.
• reject: visible heavy manipulation on creatives making product-efficacy or before/after claims.

Automation recipe

• Campaign agent generates or selects ad creatives.
• Score each variant with intended_use=publish.
• Allow-tagged creatives proceed to platform submission.
• Medium-risk creatives route to compliance review or get auto-tagged with AI-disclosure metadata before submission.
• Track which creatives passed/failed platform review and feed back into the generator's brand-safety constraint.

Evidence spans agents should inspect

• 'synthetic_subject' — generated humans or products in 'realistic' ad contexts that would require disclosure
• 'unrealistic_outcome' — before/after staging where the 'after' state appears generated
• 'celebrity_likeness' — generated content that resembles a public figure (high-risk on Meta, banned on multiple platforms)
• 'platform_policy_trigger' — visual patterns historically associated with platform-policy rejection (over-airbrushed beauty results, weight-loss before/afters, financial guarantees)

Policy pseudocode

if (result.recommended_action === "allow") continueWorkflow();
if (result.recommended_action === "revise") rewriteWith(result.evidence, result.recommended_fixes);
if (result.recommended_action === "human_review") queueForHumanReview(result);
if (result.recommended_action === "reject") discardOrRebuild();

KPIs to track

• ad disapproval rate at platform submission
• compliance-review hours per campaign launch
• FTC-disclosure compliance audit score
• % of creatives auto-tagged for AI disclosure on platforms requiring it
• time-to-launch from creative finalization to campaign go-live

What can go wrong

• Platform AI-disclosure rules evolve quickly (Meta's disclosure update in 2024, Google's expansion in 2025). Keep your policy module updated; the gate is a signal, not a substitute for current policy review.
• FTC guidance on AI in advertising is partially developed and varies by state. Pair the score with explicit legal/compliance sign-off for regulated industries.
• Some AI generation is explicitly permitted (and disclosed) and performs well in market. The gate's job is to surface, not to reject.

Cost and latency notes

Image analysis is a flat $0.02 per image. The endpoint accepts HTTPS image URLs, stores no image bytes, and logs only a URL hash plus hostname. Current v0.1 latency is vision-model-bound, so preflight balance and retry carefully.