Art.101

Article 101 – Fines for Providers of General-Purpose AI Models

Penalties SME Relevant ~2 min read · 453 words

Article 101 establishes specific fine provisions for providers of general-purpose AI models — complementing Article 99’s general penalty framework with GPAI-specific provisions. It covers both substantive violations of GPAI obligations and procedural violations such as failure to respond to AI Office information requests, with dedicated fine levels reflecting the scale of frontier AI companies.

✓ Low compliance impact for SMEs

WHAT THE ARTICLE IS ABOUT

The specific fine framework for GPAI model providers

Article 101 establishes the penalties applicable to providers of GPAI models for violations of their Chapter V obligations and for non-cooperation with AI Office enforcement proceedings. It is the enforcement counterpart to Articles 53 and 55 — specifying the financial consequences of failing to meet GPAI obligations.

WHAT IT SAYS

Up to 3% of global turnover for GPAI violations, 1% for non-cooperation

Providers of GPAI models who violate Chapter V obligations are subject to fines of up to EUR 15,000,000 or 3% of their total worldwide annual turnover for the preceding financial year, whichever is higher
Providers who supply incorrect, incomplete or misleading information to the AI Office, or who fail to respond to requests under Article 91, are subject to fines of up to EUR 7,500,000 or 1% of total worldwide annual turnover, whichever is higher
Where a provider has already been found in violation of the same provision within the preceding two years, the Commission may impose periodic penalty payments to compel compliance
Periodic penalty payments may be up to 3% of the average daily worldwide turnover for each day of non-compliance
The AI Office investigates violations and recommends fines — the Commission formally imposes them
The same mitigating factors as Article 99 apply — intentionality, cooperation, corrective measures taken, and prior violations

WHO IS AFFECTED

GPAI model providers — primarily frontier AI companies

All GPAI model providers — OpenAI, Anthropic, Google, Meta, Mistral and similar companies with models used in the EU
Open-source GPAI model providers where their models are classified as systemic risk
The AI Office which investigates and recommends fines
The European Commission which formally imposes fines
Downstream businesses using models from penalised providers — enforcement actions may affect model availability

WHAT IT MEANS FOR SMES

GPAI fines affect vendor risk — and the daily penalty rate is severe

Direct impact is minimal for most SMEs — this article targets frontier AI companies, not typical small businesses
The vendor risk angle is significant: if a GPAI provider you rely on is subject to periodic penalty payments, they are under significant operational pressure that may affect the model’s availability or terms
The 3% of daily turnover periodic penalty is particularly severe — for a company with EUR 10 billion annual turnover, this is approximately EUR 822,000 per day; this creates strong incentives for GPAI providers to comply promptly with AI Office requests
Monitor AI Office fine decisions and enforcement actions against GPAI providers — these are published and directly affect the risk profile of the AI services you depend on

Article 53 — GPAI obligations (the substantive obligations whose violation triggers these fines)
Article 55 — Systemic risk obligations (also subject to these fines)
Article 91 — Power to request documentation (non-compliance with requests triggers 1% fines)
Article 99 — Penalties (the general penalty framework this article complements)

← Back to the Guide