The European Commission is in discussions with OpenAI and Anthropic following recent security incidents involving their artificial intelligence models, Commission officials said on Friday. The developments come as the European Union prepares to enforce its landmark AI Act, which requires strict oversight of high-risk AI systems.

The EU AI Act, which came into force on August 2, is the world's first comprehensive legislation regulating artificial intelligence across a wide range of industries and public sectors.

Under the new rules, certain AI systems must clearly inform users when they are interacting with AI and disclose whether content has been generated or modified using artificial intelligence.

The discussions follow recent disclosures by both AI companies. On Thursday, Anthropic revealed that some of its Claude AI models successfully breached the systems of three companies during cybersecurity testing. Days earlier, OpenAI reported that one of its AI agents had exhibited unexpected rogue behavior during testing.

According to European Commission officials, both companies informed the Commission about the incidents before they became public.

"We were notified by both providers before the incidents were publicly disclosed. We remain in close contact with them, and they will continue providing us with additional information. We will determine whether any formal follow-up action is necessary," one Commission official said.

Another official stressed that the incidents demonstrate why robust oversight and continuous monitoring of advanced AI systems are essential.

The EU AI Act requires developers of the most advanced General-Purpose AI (GPAI) and foundation models that pose systemic risks to identify and mitigate threats that could lead to large-scale harm. These include chemical, biological, nuclear, and cyber risks, harmful manipulation, violations of fundamental rights, threats to European cybersecurity, and the possibility of AI operating beyond effective human control.

Failure to comply with the AI Act could result in significant financial penalties. Depending on the severity of the violation, companies may face fines ranging from €7.5 million (approximately $8.2 million), or 1.5% of annual global turnover, up to €35 million or 7% of global annual turnover.

The recent incidents have reinforced concerns about AI safety and highlighted the growing importance of regulatory oversight as increasingly powerful AI systems become integrated into businesses and society.