Sentinel
AI-governance & export-control readiness
Law as-of 2026-07 · verify

Governance readiness for an AI / autonomous-system deployment — not the model’s accuracy. Describe what the system does, the data and autonomy, the sectors and the jurisdictions of build and use. Sentinel returns an EU AI Act risk class, a dual-use / export-control screen, the human-authority and auditability controls you need, a clear verdict, and an exportable audit-ready control record.

Legal-tech · EU AI Act · Dual-Use Reg. 2021/821 · Wassenaar

Can you deploy this AI system — and can you prove it was governed?

Sentinel fuses the two questions a regulated AI programme has to answer at once: is it lawful to build and use (EU AI Act risk tier and the obligations it triggers), and is it lawful to move (dual-use export controls on surveillance, intrusion, cryptography and advanced compute). It then names the meaningful-human-control and logging / auditability measures that make the answer defensible — regulated by design: human authority, auditability, export screening.

Decision-support, not legal advice. The AI Act timeline is moving: prohibitions apply from 2 Feb 2025, GPAI duties from 2 Aug 2025, Art 50 transparency from 2 Aug 2026, and the Digital Omnibus (adopted by the European Parliament 16 Jun and Council 29 Jun 2026, final act signed 8 Jul 2026) moves Annex III high-risk obligations to 2 Dec 2027 and adds a new Art 5 prohibition on AI that generates non-consensual intimate imagery or CSAM (from 2 Dec 2026) — adopted but not yet in force: awaiting Official-Journal publication (then in force three days later), so confirm before you rely on it. Control lists and sanctions change too. Positions are dated July 2026 and labelled legal fact vs reasoned assessment; confirm current sources before you rely on them.

Describe the deployment

9 inputs
Load an illustrative worked example
4 · Sensitivity & context
8 · Jurisdictions
9 · Dual-use / security context — be honest; this drives export screening
Your readiness assessment will appear here
Describe the deployment on the left (or load a worked example) and press Assess the deployment. Sentinel returns the AI Act risk class, the export screen, the human-control & logging measures, a verdict and an exportable record.
Sentinel — built by Christopher Farmer · runs fully offline, no network calls · decision-support tool. Project 6 · portfolio