EU Gains Access to Anthropic’s Mythos AI Model After Months of Negotiations

The European Union’s cybersecurity agency has gained access to Anthropic’s highly capable Mythos AI model after months of negotiations involving the company, European officials and the U.S. government, marking a significant step in Europe’s effort to evaluate frontier artificial intelligence with advanced cybersecurity capabilities.

The development was reported by Coin Bureau and follows concerns in Europe over limited access to Mythos after Anthropic initially restricted the model to a small group of organizations. The European Union had sought access partly because of the model’s potential importance to cybersecurity research and defense.

Anthropic’s Claude Mythos was introduced in April as part of Project Glasswing, a program designed to use advanced AI to identify vulnerabilities in critical software before similarly capable systems become more widely available. Anthropic said its initial partners had used Mythos to identify more than 10,000 high- or critical-severity vulnerabilities across important software projects.

U.S. Export Controls Complicated European Access

Access to Mythos became more complicated after the U.S. government imposed restrictions on the model in June.

Anthropic said the U.S. government had ordered it to suspend access to Mythos 5 and Fable 5 for foreign nationals, citing national-security authorities. The company subsequently said access to Mythos was restored for a limited group of U.S. organizations after receiving government approval.

The restrictions underscored the growing overlap between advanced AI development, cybersecurity and national security. Mythos is particularly sensitive because its capabilities extend beyond conventional software assistance into identifying and developing potential exploits.

European officials had already raised concerns about the strategic implications of being unable to evaluate the technology directly. A European markets Parliament question submitted in April highlighted the cybersecurity risks associated with the model and called for a stronger European response to rapidly advancing AI capabilities.

EU Cybersecurity Agency Begins Mythos Testing

The European Commission said on Thursday that the EU’s cybersecurity agency, ENISA, has been granted access to Anthropic’s Mythos 5. The agency is testing the model to assess its capabilities and potential implications for cybersecurity.

The access comes as governments increasingly seek direct experience with advanced AI systems rather than relying solely on information supplied by developers. For Europe, testing Mythos could provide cybersecurity authorities with a clearer markets assessment of both its defensive applications and the risks posed by increasingly capable AI-driven vulnerability discovery.

The immediate focus will be how ENISA evaluates Mythos in controlled environments and whether the results influence Europe’s approach to governing advanced AI systems with significant cyber capabilities.

Writer: Victoria Hale
Technology & Blockchain Writer

Victoria Hale writes about blockchain technology, digital infrastructure, and the intersection of emerging technologies with finance. Her articles explore how new protocols and systems are shaping the evolving digital economy.

She prioritises clarity and accuracy when explaining technical developments to a general audience.