EU AI Act Enforcement Begins: Article 50 Transparency and GPAI Powers Now Live

The European Union's AI Act reached its most consequential enforcement milestone on August 2, 2026, when the European Commission's AI Office and member-state authorities gained full penalty powers over general-purpose AI providers and Article 50 transparency obligations became binding for most AI systems deployed in the EU, the Commission's digital strategy site and beam.ai confirm. Anthropic and OpenAI, as the two largest GPAI providers headquartered outside the EU, are squarely in the regulator's crosshairs.

What Activated on August 2

Three things changed on August 2. First, the Commission gained penalty enforcement powers over GPAI providers, with fines up to 3 percent of global turnover or €15 million for the most serious violations. Second, Article 50's transparency obligations activated, requiring AI-generated content to be clearly labeled and chatbot users to be informed they are talking to an AI. Third, national market-surveillance authorities in every EU member state can now fully investigate and sanction AI Act breaches. High-risk Annex III systems — a separate category that triggered much of the pre-August worry — were pushed to 2027-2028 under the May 7 Digital Omnibus on AI amendment.

Why Anthropic and OpenAI Are Most Exposed

The GPAI Code of Practice split the model market into providers who signed and providers who did not, and that split is now a procurement signal. Anthropic and OpenAI are both names widely cited by European enterprises as critical infrastructure, but both have reportedly declined to sign the full Code, exposing them to maximum scrutiny. "The August 2 deadline is the moment GPAI becomes a regulated industry, not a self-regulated one," said Lucilla Sioli, director of the Commission's AI Office. Anthropic separately faces use-of-system-data compliance questions around its Claude training, while OpenAI's new GPT-5.6 Sol with 750-token/second Ultrafast mode is being examined for compute-disclosure compliance.

California and the U.S. Backing Layer

Across the Atlantic, the policy pressure is multiplicative. Senator Jim Banks' August 14 letter urged the White House to limit Chinese open-weight AI proliferation, while California's SB 53 — the first U.S. frontier-model safety law — has been in force since January 1, 2026. Two California AI bills — focused on chatbot safety for children and AI copyright transparency — face their final Senate and Assembly votes on August 13. Together, the EU AI Act, California's SB 53, and emerging federal action mark the first year in which frontier-model developers must operate under binding safety law in three major jurisdictions simultaneously.