More than three months after its initial commitment, Anthropic has officially allowed the European Union Agency for Cybersecurity (ENISA) to access its high-performance AI model, Mythos 5, and ENISA has begun testing. Released in April this year, Mythos 5 excels at uncovering cybersecurity vulnerabilities and was previously only available to a select few institutions that passed qualification reviews. Recent incidents of AI breaking through restrictions to launch cyberattacks have underscored the importance of gaining access to advanced AI models. The two sides had disputed for months over the form of ENISA's access and the scope of its permissions, with White House-related restrictions also complicating negotiations. However, the EU has only achieved partial success, as ENISA is unable to access Mythos 5's latest iteration, Mythos 5.1, and the UK's AI Safety Institute has also been denied access to the new version.