EU begins enforcing AI Act, first RFIs to model providers
anthropic google openai startup
| Source: HN | Original article
The European Union has started enforcing the AI Act, issuing its first requests for information to AI model providers.
The European Union has moved from planning to policing its new AI legislation. On 29 August 2026 the EU’s AI Office issued its first formal requests for information (RFIs) to a handful of leading general‑purpose AI (GPAI) model providers, including OpenAI, Anthropic and Google. The RFIs are part of the enforcement regime that came into force on 2 August 2026, when the AI Office and national authorities were given the mandate to supervise compliance with the AI Act.
The move signals that the EU is now actively testing whether the most widely used foundation models meet the Act’s transparency, risk‑assessment and safety obligations. Under the framework, the AI Office can issue “simple” RFIs directly; if a provider’s reply is inaccurate or misleading, the office may levy fines. More formal RFIs can be issued by the European Commission, expanding the toolbox for regulators. By targeting GPAI models—the core engines behind chatbots, code assistants and AI agents—the EU is focusing on the technology that underpins a large share of commercial AI services.
For the industry, the RFIs are a wake‑up call. Providers must be ready to disclose technical documentation, training data provenance and risk‑mitigation measures that the Act requires. Failure to comply could result in monetary penalties and, in extreme cases, restrictions on market access across the bloc. The scrutiny also raises questions about how quickly companies can adapt their internal compliance processes, especially given the rapid pace of model updates.
What to watch next: how the named firms respond to the EU’s queries, whether the AI Office escalates to formal Commission‑issued RFIs, and the potential issuance of fines for non‑compliant answers. Observers will also monitor whether the enforcement wave expands to smaller model developers and to downstream AI applications that embed GPAI models. The next few weeks will reveal how the EU’s regulatory muscle reshapes the European AI market and whether it prompts broader shifts in global AI governance.
Sources
Back to AIPULSEN