RedHub AI
How the roll-up gate works
How the AI Disclosure & Labeling gate works
This tool grades your AI surfaces against EU AI Act Article 50 transparency obligations. It reads 6 AI surfaces and rolls them into one verdict — the worst, never the average — then a gate that can only make it worse.
The 6 AI surfaces
A synthetic surface presented as real with no machine-readable mark — a deepfake ad with no label — is deceptive, and one deceptive surface makes the whole set NOT COMPLIANT however clean the rest is. Undisclosed synthetic content passed off as real is the exact harm Article 50 targets.
Why: Four of six surfaces are clear and one is only partly marked — but the spokesperson "deepfake" ad is presented as real with no machine-readable mark and no deepfake label. That's a deceptive synthetic surface, and one deceptive surface makes the whole set NOT COMPLIANT however clean the rest is: undisclosed synthetic content passed off as real is exactly what Article 50 targets.
Most surfaces can be labeled — but one synthetic surface passed off as real with no disclosure is dispositive: an undisclosed deepfake ad sinks the whole set. The roll-up takes the worst surface, the deceptive-surface gate forces NOT COMPLIANT regardless of the score, and it names the one surface to fix first. A pre-publish readiness audit for EU AI Act Article 50 transparency obligations — not a conformity assessment and not legal advice. "NOT COMPLIANT" flags an unmet Article 50 disclosure obligation in the inputs you gave, not an overall legal conclusion about your organization. The Act's text and timelines are still moving — confirm current obligations with qualified counsel.