Anthropic has launched Project Glasswing, a cybersecurity initiative built around its unreleased model Claude Mythos Preview and a group of major technology and infrastructure partners. The company says the goal is to help secure “the world’s most critical software” at a time when AI tools are becoming powerful enough to find and exploit software flaws at scale. Anthropic said the project includes Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks.
About the Project Glasswing
Project Glasswing is not a public product launch, it is a restricted cybersecurity program in which select organizations can use Claude Mythos Preview for defensive security work. Anthropic said the model has already identified thousands of high-severity vulnerabilities, including flaws in major operating systems and web browsers, and that the project is meant to turn those capabilities toward defense before they spread to hostile actors. Reuters reported the same broad purpose and said Anthropic is trying to prepare the industry for AI-driven cyber risk.
The launch partners span cloud computing, software, cybersecurity, finance, and open-source infrastructure. Anthropic’s official release names AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. Amazon, Microsoft, Apple, Google, CrowdStrike, Palo Alto Networks, and Nvidia are key partners.
Anthropic said the partners will use Mythos Preview in their own defensive security work and will share findings across the industry. In the official announcement, Cisco said AI has crossed a threshold that changes the urgency of protecting critical infrastructure, AWS said it has already been testing the model in its security operations, Microsoft said the model can help identify and mitigate risk earlier, and CrowdStrike said defenders need to move as fast as attackers. Google said it will make the model available through Vertex AI, while the Linux Foundation said the project could help open-source maintainers find and fix vulnerabilities at scale.
Anthropic is extending access to more than 40 additional organizations that build or maintain critical software infrastructure. The company is committing up to $100 million in usage credits and $4 million in direct donations to open-source security groups.
The objective is to give defenders an edge in a cybersecurity environment where AI models can now do more than assist with simple coding tasks. Anthropic said the broader aim is for users to “safely deploy Mythos-class models at scale,” The company is trying to make those capabilities useful for defensive cybersecurity work rather than harmful use.
That urgency is the real story behind Project Glasswing. Anthropic said the model has found thousands of vulnerabilities across operating systems and browsers, and the cybersecurity industry is already under pressure from AI-powered attack techniques. In practical terms, the project is meant to shorten the gap between vulnerability discovery and patching, especially for systems that support finance, communications, logistics, health, and government operations.
Project Glasswing is Anthropic’s attempt to move advanced AI into a controlled cybersecurity setting, with big technology companies and infrastructure providers testing how these models can help fix software before attackers do.
[ Also Read: Microsoft, NVIDIA and Anthropic Announce Strategic Partnerships to Advance AI Infrastructure ]
