Pasquale Pillitteri reported that the European Commission published draft implementing regulation Ares(2026)9589403 on October 12, 2026, setting out proposed procedures for investigations and penalties under Articles 75c and 75d of the EU AI Act. The draft is not the source of the AI Office’s enforcement powers: Regulation (EU) 2026/1744 is already in force and establishes those powers within a defined remit.
Which AI systems fall within the AI Office’s remit?
The AI Office’s direct supervision covers specified systems: certain AI systems built on general-purpose AI models when the model and system were developed by the same provider or by providers within the same undertaking, subject to statutory exceptions; and systems that constitute or are integrated into designated very large online platforms or search engines. The rules also cover deployers when they are themselves the provider or belong to the same undertaking. Ordinary use of a third-party chatbot, by itself, does not put a business under the AI Office’s direct supervision.
The enacted framework is set out in Regulation (EU) 2026/1744. It gives the AI Office powers to request information, conduct remote or on-site inspections, question people, seal premises or records, order access to AI systems, require data and document preservation, and appoint experts or auditors. The Office can also make non-compliance findings and impose fines and periodic penalty payments within its remit.
Procedures described in the Commission draft
Pillitteri reports that the draft would allow access to the case file after preliminary findings. The account says documents containing third-party trade secrets would be redacted, while unredacted material would be restricted to named external advisers and experts.
Pillitteri also reports a proposed five-year limitation period for imposing fines, starting from an infringement or the end of continuing or repeated conduct. Investigative acts would restart the period, subject to a ten-year cap and suspension periods. These are reported provisions of the draft, not procedural rules established by Regulation (EU) 2026/1744.
Fine ceilings and daily penalty payments
Pillitteri reports three existing AI Act fine ceilings associated with the proposal. For large companies, the account says the higher of the fixed amount or the turnover percentage applies; for SMEs and small mid-caps, the lower amount applies.
| Violation category | Reported fine ceiling |
| Prohibited practices under Article 5 | €35 million or 7% of revenue |
| Other violations, or failure to comply with decisions or commitments | €15 million or 3% of revenue |
| Inaccurate, incomplete or misleading information | €7.5 million or 1% of revenue |
The enacted law sets a separate limit for periodic penalty payments: up to 5% of average daily income or worldwide annual turnover in the preceding financial year for each day. That per-day ceiling is distinct from the reported fine tiers.
Consultation deadline
Pillitteri reports that the consultation closes on November 9, 2026, at 11:59 p.m. Brussels time. The European Commission’s Have Your Say page for initiative 19292 is the consultation portal.
