Anthropic Unveils New AI Cybersecurity Project

Anthropic has launched a new cybersecurity-focused initiative, partnering with major technology companies including Amazon, Microsoft, and Apple to advance AI-driven defense capabilities.

The program, called “Project Glasswing,” will give select organizations early access to an unreleased AI model named Claude Mythos Preview. According to Anthropic, the model is designed to support defensive cybersecurity efforts by identifying vulnerabilities and strengthening digital infrastructure.

Additional partners in the initiative include CrowdStrike, Palo Alto Networks, Google, and Nvidia.

The announcement comes amid growing concerns about AI-driven cyber threats. At this year’s RSA Conference, industry leaders widely discussed the increasing sophistication of AI-powered attacks and the need for more advanced defenses.

In a recent blog post, Anthropic said the Mythos Preview model has already identified thousands of significant vulnerabilities across operating systems, web browsers, and other software platforms. Insights gathered through the initiative will be shared with the broader industry to improve collective security.

The company is also expanding access to around 40 additional organizations responsible for critical software infrastructure. As part of its commitment, Anthropic has pledged up to $100 million in usage credits and $4 million in funding for open-source security projects.

Anthropic stated that its long-term goal is to enable safe, large-scale deployment of advanced AI models like Mythos. The company also confirmed it is in ongoing discussions with the U.S. government regarding the model’s capabilities and potential risks.

The urgency around AI security continues to grow. Anthropic revealed that hackers previously exploited vulnerabilities in its Claude AI system to target about 30 global organizations. Separately, a study by IBM and Palo Alto Networks found that 67% of surveyed executives experienced AI-related cyberattacks in the past year.