Restricted · Internal
demo-token
Every adapted asset is decoded, sampled and checked against real food-advertising and AI-disclosure rules before it ships — HFSS, nutrition and green claims, kids'-audience limits, dairy-designation and AI-provenance law. Fails and flags come back with the exact instrument that fired and a one-line fix, routed straight into MRCL MAKE.
60-second orientation
Pick an asset, run the content scanner, pick your markets, and the checker runs the rule engine over every selected market — returning a per-market verdict for each rule that fires, worst-first, with the real legal instrument named and a one-line fix.
Every rule resolves to one of three verdicts. A market's overall standing is driven by its worst rule.
A hard restriction is triggered — the asset cannot run as-is in this market/placement. Needs a change before it ships.
A conditional or disclosure obligation applies — usually clearable with a label, substantiation, or a placement tweak.
No specific additional restriction found in the illustrative ruleset for this asset in this market.
Rules fire on scanner-extracted attributes — visible claims found by OCR/text heuristics (e.g. nutrition_claim, vegan_claim, sustainability_claim, kids_appeal), frame facts such as duration/dimensions, the declared AI share where available, and the channel (digital / TV / social / OOH). The claim toggles remain visible as a manual override for variant testing.
The ruleset covers, among others: UK HFSS advertising restrictions (Oct 2025), the EU Nutrition & Health Claims Regulation (EC) 1924/2006, EU AI Act Art. 50 transparency, Chile Ley 20.606 / Mexico NOM-051 front-of-pack & minors regimes, kids'-advertising codes (UK CAP/BCAP, Nordic broadcast bans), Halal substantiation expectations, EU dairy-designation rules (Reg 1308/2013), green-claims law (EU Dir. 2024/825 / UK CMA Green Claims Code) and US FTC truth-in-advertising. Every instrument named is real; the mapping to this asset is illustrative.
Two inputs drive every verdict. The rulebook: each rule in this build was drafted directly from the primary legal texts — legislation.gov.uk (Communications Act 2003 s.321A–C as amended, the UK HFSS regime), EUR-Lex (Reg (EC) 1924/2006 nutrition & health claims; Reg (EU) 1308/2013 dairy designations; Reg (EU) 2024/1689 — the AI Act, Art. 50; Dir (EU) 2024/825 green claims), Chile's Ley 20.606 (MINSAL), Mexico's NOM-051 (DOF), Peru's Ley 30021, the UK CAP/BCAP codes (ASA), Nordic broadcast statutes, halal standards (BPJPH Indonesia · JAKIM Malaysia · GSO 2055-1 Gulf), the UK CMA Green Claims Code, and the US FTC Act §5 with the FTC's advertising guides. The asset: this build now decodes the selected supplied MP4 in the browser, samples real frames to canvas, OCRs visible copy when the OCR worker is available, and derives claim tags from that content. Channel, placement and AI-provenance still come from asset delivery metadata because those are not fully inferable from pixels alone. In production the rulebook is a counsel-maintained rules service with monitored regulatory updates.
Every finding has a Why? button opening the full rule text, exactly what in this asset triggered it, and what would clear it. Fail and Flag rows also get Send fix to MRCL MAKE ▸ — it drafts a one-line remediation brief (asset · market · rule · required change) and hands it into the MRCL MAKE Human x AI craft system, mirroring the Occasionist handoff.
The regulation names and rule summaries are genuinely accurate. The scanner pass is real browser-side media analysis, but OCR quality depends on frame resolution, motion, crop and browser support. The legal rulebook is still illustrative and pending Publicis legal/regulatory sign-off. See the note at the foot of the page.