Anthropic and Claude Mythos: The AI Challenging Global Cybersecurity

Anthropic, an artificial intelligence developer, recently unveiled its latest model, Claude Mythos, which has proven so effective in detecting cybersecurity vulnerabilities that it will not be made public. The company has decided to provide this tool exclusively to major technology infrastructure providers so they can address the identified flaws.

Since the end of March, rumors have circulated about this AI model named Claude Mythos. Anthropic has now confirmed that Mythos poses a potential threat to cybersecurity, as it is capable of detecting and exploiting online vulnerabilities. According to the company, AI models have reached a level of coding proficiency that allows them to outperform nearly all human experts in this field.

Anthropic revealed that Mythos Preview, which is not publicly accessible, has already identified thousands of serious vulnerabilities in every major operating system and web browser. When asked, a representative from Anthropic referred to a company blog post for more details.

To counter these risks, Anthropic launched Project Glasswing, a consortium that includes companies such as Apple, Amazon Web Services, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, Nvidia, and Palo Alto Networks. These organizations, along with more than 40 others, will have access to Mythos to strengthen their defenses against AI-based attacks. Anthropic is committed to providing $100 million in usage credits for Mythos and $4 million in donations to open-source security organizations.

Dario Amodei, CEO of Anthropic, expressed on X that while the dangers are evident, there is a real opportunity to make the internet and the world safer through AI-powered cybersecurity capabilities. In a YouTube video about Project Glasswing, leaders from Microsoft, the Linux Foundation, and Anthropic discussed the potential damage caused by software vulnerabilities. Cloud computing companies have already collaborated with the new model to identify flaws. Anthony Grieco, Cisco's security and trust officer, wrote in a blog post that AI-powered analysis uncovers data at a scale and depth that legacy frameworks were not designed to accommodate.

Amazon Web Services also noted that the model has found ways to strengthen code even in its most proven systems. Amy Herzog, vice president and chief information security officer at AWS, described Mythos Preview as a radical shift in reasoning and AI capabilities for cybersecurity.

The phenomenon of AI discovering and potentially exploiting software vulnerabilities is not new. The DARPA Cyber Grand Challenge has already seen several instances of AI drawing attention in this area, said Michal Salát, director of threat intelligence at Norton, the antivirus provider. However, AI technology that could be accessible to everyone possesses some of these capabilities. Anthropic's Project Glasswing is focused on safeguarding this powerful technology, which can transform vulnerability research but also pose a serious risk if misused for malicious purposes.

While it represents a significant advancement over current models such as Opus 4.6, the underlying capability already exists today, and vulnerability research is rapidly emerging as one of the primary concrete use cases for AI in cybersecurity.

National policymakers, who have fluctuated between the need for federal AI regulation, will likely closely monitor the consortium's progress. Senator Mark Warner praised the initiative in a statement. "I applaud these leading companies for recognizing this threat and proactively sharing information, capabilities, and computing resources to better protect our critical infrastructure," said the Virginia Democrat. "As AI significantly accelerates the discovery of new vulnerabilities, I hope the industry will accelerate and reprioritize patching accordingly." Warner, whose state is home to AI data centers, recently called a proposal for a moratorium on building data centers "nonsense," but also warned of the societal risks posed by the rapid development of AI, leading to massive job losses.