Anthropic, the artificial intelligence company widely regarded as the main rival to OpenAI's ChatGPT, is broadening the rollout of its powerful cybersecurity model, Mythos. The company announced it is adding approximately 150 new partners across more than 15 countries, including organisations in the power, water, healthcare, communications, and hardware sectors.
Mythos was first unveiled in April under the name "Project Glasswing," initially limited to 50 partners. According to Anthropic, those early testers have already identified more than 10,000 high-severity or critical security vulnerabilities, including flaws in every major operating system and web browser. The model is designed to detect weaknesses in codebases before malicious actors can exploit them.
The expansion comes with a stark warning. Anthropic estimates that for most of its new partner organisations, a successful cyberattack on their systems could affect more than 100 million people. "What each partner has in common is that a successful attack on their codebase could be catastrophic," the company stated.
European Institutions in the Mix
One potential partner is the European Commission. US media outlets have reported that Anthropic has approached the EU's executive body to offer access to Mythos. Euronews Next is working to independently confirm this contact. If confirmed, it would mark a significant step in the deployment of advanced AI for cybersecurity within European institutions, which have been increasingly focused on digital resilience.
The expansion also raises questions about the broader geopolitical context. As the US considers expanding nuclear bomber deployments in Eastern Europe, the security of critical infrastructure across the continent becomes even more pressing. Mythos could play a role in hardening systems against state-sponsored attacks, though its own vulnerabilities remain a concern.
Anthropic's decision to widen access is partly driven by a fear of what lies ahead. The company predicts that within the next year, more AI firms will develop similarly powerful models and "could release them without safeguards that prevent misuse." In that scenario, the company warns, "cyberattacks could occur much more often, and in much more unpredictable forms."
This proactive stance mirrors broader trends in AI governance. The White House has offered voluntary AI model vetting after a security incident involving Anthropic, underscoring the delicate balance between innovation and safety. Meanwhile, Microsoft has unveiled its own in-house AI models, reducing its reliance on both OpenAI and Anthropic, which could reshape the competitive landscape.
For European readers, the expansion of Mythos is not just a tech story. It touches on the security of energy grids, hospitals, and communication networks across the continent. As Anthropic brings more partners on board, the question remains whether the safeguards are sufficient to prevent the very catastrophe the company warns about.
