Home · EU AI Act · Articles · Article 55
EU AI Act · Engineering reference

Article 55 — GPAI Systemic-Risk Obligations

GPAI models above the 10^25 FLOP training-compute threshold carry additional systemic-risk obligations including evaluations, adversarial testing, incident reporting and cybersecurity.

Audit-survivable
18%
of 50 EU-exposed enterprises in MindMap's 2026 readiness benchmark

What the Article requires

Article 55 adds obligations on top of Article 53 for GPAI models with systemic risk — defined as models trained on >= 10^25 FLOPs of compute, or otherwise designated by the AI Office. Obligations: state-of-the-art model evaluations including adversarial testing, assessment and mitigation of systemic risks at Union level, serious-incident tracking and reporting, and adequate cybersecurity protection for the model and its physical infrastructure.

In engineering terms

Article 55 affects very few enterprises directly — the threshold catches frontier-model providers like OpenAI, Anthropic, Google, Mistral. For enterprise integrators the question is which downstream evidence each frontier vendor provides under Article 55 — adversarial-test reports, systemic-risk assessments, incident logs. The vendor disclosures cascade into customer-side Articles 9-15 evidence.

Compliance checklist

  • ✓Confirm whether your GPAI provider crosses the 10^25 FLOP threshold
  • ✓Obtain vendor Article 55 disclosure pack
  • ✓Cascade vendor Article 55 evidence into customer Article 11 documentation
  • ✓Adversarial-test review (vendor-provided)
  • ✓Incident-tracking integration if vendor-managed

Terms used here

GPAI (General-Purpose AI)EU AI ActAnnex III high-risk AI

All Articles in the reference · The EU AI Act compliance architecture

Free download · 12-page fill-in PDF

Get the Article 55 checklist as a fill-in template

The EU AI Act Readiness Kit puts this checklist, and the other 13 Articles', into one fill-in PDF with room for an owner, evidence location and target date on every item. Plus an AI system inventory, a classification record, an Annex IV outline and a FRIA template.

  • 01How to use this kit. Four steps, and which checklists apply to each classification result.
  • 02AI system inventory. One row per AI system: owner, what it does, built or bought, your role and its classification.
  • 03Classification record. The five questions from our applicability checker, with space for answers, notes and the result.
  • 04Evidence checklists. 72 evidence items across Articles 9 to 15, 25, 26, 27, 50, 53, 55 and 72, each with an owner, evidence location and target date.
Get the kit
Free PDF, in your inbox in 30 seconds.
We'll send the PDF, and may follow up once or twice with related material. One click unsubscribes.

Need audit-survivable evidence for Article 55?

MindMap runs a 90-day path from standing start to audit-survivable evidence. Talk to the engineering team.

Download the whitepaper →Check your tier →Book a scoping call →