0 days until EU AI Act high-risk enforcement (2026-08-02 UTC)

If your AI product touches the EU, you have 0 days to be compliant.

EU AI Act high-risk obligations begin August 2, 2026. Penalties for high-risk non-compliance reach €15M or 3% of global revenue (Art. 99). EpochCore ships you Article 12 record-keeping in days, not quarters.

Not legal advice. This page summarizes public guidance from the EU AI Act (Regulation (EU) 2024/1689). Your scope assessment, conformity assessment route, and DPIA still belong with your counsel. EpochCore delivers the technical substrate — tamper-evident, audit-ready record-keeping — that produces evidence supporting Articles 12, 15, and 17. Two penalty tiers exist: prohibited practices (Art. 5) reach €35M or 7%; high-risk non-compliance (Art. 99) reaches €15M or 3%. The high-risk tier is what applies to most providers placing a high-risk system on the EU market.

What the Act actually requires for high-risk AI

From Regulation (EU) 2024/1689. The high-risk obligations are not optional — they are conditions for placing a high-risk system on the EU market.

ARTICLE 9

Risk management system

Continuous, iterative risk identification, analysis, and mitigation across the system lifecycle. Documented.

ARTICLE 12

Automatic record-keeping

Automatic logging of events while in operation, capable of identifying when the system poses a risk or undergoes substantial modification. Logs must be tamper-evident.

ARTICLE 13

Transparency to deployers

Operating instructions, characteristics, capabilities, limitations — clear, complete, and verifiable.

ARTICLE 14

Human oversight

System designed to be effectively overseen by natural persons during use. Override mechanisms documented.

ARTICLE 15

Accuracy, robustness, cybersecurity

Resilience against errors, faults, inconsistencies. Cybersecurity measures resilient to attempts to alter use or performance.

ARTICLE 17

Quality management system

Documented, including post-market monitoring, incident reporting, and (for providers) conformity assessment evidence.

How EpochCore delivers it

Article 12 says "tamper-evident logs." Article 15 says "cybersecurity resilient to alteration." Article 17 says "documented evidence." EpochCore produces evidence supporting all three with one D-KaP envelope per agent decision.

ArticleRequirementHow EpochCore satisfies it (as a technical control)
Art. 12Automatic, tamper-evident logging while in operationD-KaP envelope per agent output: hybrid Ed25519 + ML-DSA-65 signature, chain-anchored to a public RAS root, third-party verifier. Tamper = chain break.
Art. 13Operating instructions verifiable by deployerProduces evidence supporting deployer-facing documentation: each envelope carries operating-context metadata (model version, policy state, agent identity). Conformity-assessment route and deployer instructions remain the provider's responsibility.
Art. 14Human oversight mechanisms documentedProduces audit evidence of human-oversight events: per-envelope policy state field records whether decision was auto-issued or human-confirmed. The oversight mechanism itself is your product responsibility; we evidence it.
Art. 15Cybersecurity resilient to alterationPost-quantum cryptography (NIST FIPS 204 / ML-DSA-65) on every envelope. Resilient against today's threat model and forward-secure against future quantum adversaries.
Art. 17Quality management documented evidenceAudit-trail-keeper agent emits chain-anchored daily compliance digests. Conformity assessment evidence pack on demand.
Technical detail (for security / GRC readers)

Each envelope is a JSON object containing the SHA-256 hash of the canonical decision plus a hybrid signature: Ed25519 (classical) and ML-DSA-65 (NIST FIPS 204 post-quantum, security level 3). The envelope is chained to the previous envelope on your tenant's chain and to the public RAS root 40668c787c463ca5. Verification is open at https://chain.epochcoreqcs.com/v1/verify — no SDK required, no proprietary tooling, no vendor lock.

Pricing

Annual, prepaid. White-glove onboarding. Cancellation pro-rated to the end of the term where required by applicable law.

Starter
$39,000 / year
For mid-size AI products with EU exposure
  • 1M sealed events / month
  • Article 12 + 15 + 17 evidence pack
  • Quarterly audit-ready digest
  • Standard DPA + Privacy Policy
  • Email support
Book the compliance call →

Book a 20-minute compliance call

Tell us what your AI product does. We'll tell you which Articles apply and whether $39k or $99k is the right fit. No commitment.

Booked. We'll reply within one business day with two time options.