ActCheckIt is an EU AI Act compliance API for high-risk AI systems. One POST request returns a structured Annex IV assessment with article citations, severity verdicts, and PDF audit reports — ready for legal review.
High-risk AI systems operating in the EU must demonstrate compliance with the EU AI Act by August 2, 2026. Fines reach €35 million or 7% of global annual turnover — whichever is higher.
Every response evaluation maps directly to EU AI Act articles and Annex IV documentation requirements.
Every flag cites the specific EU AI Act article it implicates. Not general ethics — precise regulatory mapping.
Results are organised into the 9 Annex IV sections your technical documentation must address.
Download a compliance-ready PDF for every evaluation. Formatted for legal and compliance team review.
No evaluation content is stored after delivery. No legal discovery exposure. Your data stays yours.
Send your proposed AI response before delivering it to users. Get back a structured EU AI Act assessment with article citations and a recommended action.
// Request { "proposed_response": "This AI system assists doctors by flagging potential drug interactions and explains its reasoning. All final decisions remain with the licensed physician.", "context": "Medical decision support tool in EU hospitals", "risk_tier": "high" } // Response { "recommended_action": "pass", "flag_count": 0, "highest_severity": "none", "risk_tier_applied": "high", "regulation": "EU AI Act (Regulation EU 2024/1689)", "evaluation_id": "eval_a7f3c92d", "annex_iv_assessment": { /* full 9-section breakdown */ } }
{
"recommended_action": "block",
"flag_count": 1,
"highest_severity": "critical",
"flags": [
{
"dimension": "D3 — Harm by Omission",
"severity": "critical",
"eu_articles": ["Article 9", "Annex III(5)(b)"],
"annex_iv_section": 2,
"explanation": "Response omits system limitations and
false negative handling required at high-risk tier.",
"suggested_revision": "Add explicit disclosure of system
scope, known failure modes, and escalation procedures."
}
]
}
ActCheckIt evaluates the sections it can evaluate from response content — and is transparent about what requires your own documentation.
"ActCheckIt evaluates what can be evaluated from AI response content. What requires your team's documentation, we flag clearly — not silently."
Two options — pay as you go, or the full compliance tier with PDF audit reports included.
Enterprise volume pricing available. Contact us →
Compliance and ethics are not the same thing. A response can satisfy every EU AI Act requirement and still cause harm. The Works Ethic evaluates all 15 dimensions across every major ethics framework — free to start.
Try it for free →Not in the EU AI Act. Still harmful.
ActCheckIt does not store evaluation content after delivery. No response text, no context, no flags — nothing. Every evaluation is generated at call time and delivered directly to you. There is nothing on our servers to subpoena, breach, or accidentally expose. This is not a setting. It is how the system is built.