AIcleared is an interactive readiness workspace for SMEs and engineering teams: draft Article 50 disclosure, keep an Annex IV file alive, practise tamper-evident logging, and catch Article 25 traps before procurement does.
Not sure the AI Act applies to you at all? Take the 2-minute plain-language check — seven questions, no jargon.
The compliance vacuum for agile teams is widest exactly when enforcement starts.
Machine-readable marking is enforceable for new systems; systems on the market before Aug 2026 mark from 2 Dec 2026.
The AI Office coordinates national authorities; investigations and breach reports make controls verifiable, not paper policies.
€35M / 7% only for banned practices; €15M / 3% for the middle tier — and SMEs are capped at the lower of the two (Art. 99(6)).
Annex III lands 2 Dec 2027 (Annex I: 2 Aug 2028) — documentation discipline starts today.
Product suite
Enterprise GRC platforms own policy packs. AIcleared owns the acute technical bottlenecks — signing an asset, regenerating a living technical file, proving a log, or killing a silent provider reclassification.
Draft and preview C2PA-style provenance manifests and EU disclosure labels for AI-generated media. Preview only — no files are signed here.
Turn high-risk technical files into a continuously updated evidence pack mapped to the Annex IV points — export ready for authority requests at any time.
Tamper-evident agent traces in a SHA-256 hash chain that preserve auditability after GDPR erasure, using keyed PII tags.
Detects white-label, modification, and purpose-shift triggers that can reclassify deployers as providers — correctly scoped to high-risk systems.
Regulatory timeline
Feb 2025
Unacceptable-risk practices banned; AI literacy duties begin. The Digital Omnibus added two further prohibitions — non-consensual intimate imagery and child sexual abuse material — applicable from 2 Dec 2026.
Aug 2025
General-purpose AI provider duties apply. The AI Office is operational; several Member States missed the deadline to designate their competent authorities.
Aug 2026
Article 50 duties for new systems; broad enforcement & whistleblower coverage.
2 Dec 2026
Art. 50(2) marking applies to systems placed on the market before Aug 2026 (Digital Omnibus).
2027–28
Annex III standalone systems: 2 Dec 2027. Annex I embedded systems: 2 Aug 2028.
Dates reflect the Digital Omnibus on AI (European Council final adoption 29 June 2026), which deferred the high-risk deadlines above and added a prohibition on generating non-consensual intimate imagery.
Article 99
Middle-tier failures — missing documentation, weak oversight, transparency gaps — carry up to €15M or 3% of worldwide turnover; banned practices up to €35M or 7%. For SMEs and start-ups every tier is capped at the lower of the fixed amount or the percentage (Art. 99(6)) — the exposure calculator applies both rules.
€35M or 7%
€15M or 3%
€7.5M or 1%