Anthropic is introducing a new AI model as part of a cybersecurity partnership with Nvidia, Google, Amazon Web Services, Apple, Microsoft and other companies. Project Glasswing, as it’s called, is supposed to be a way for large companies, and possibly even the government, to flag vulnerabilities in their systems without any human intervention.
Anthropic is offering its launch partners access to the Cloud Mythos Preview, a new general-purpose model that it is not currently planning to release publicly due to security concerns. Newton Cheng, cyber lead of Anthropic’s Frontier Red Team, said The Verge This model would ideally give cyber defenders a “head start” against adversaries. Partners will use the model to analyze their systems to detect high-risk vulnerabilities and help fix them. Access is restricted to prevent the same adversaries from finding weak points and using it to launch attacks.
Although Cloud Mythos Preview was not trained specifically for cybersecurity purposes, Anthropic said in a release that the model’s “strong agentic coding and reasoning skills” are behind its cybersecurity advancements. in an interview with The VergeNewton Cheng, cyber lead for Anthropic’s Frontier Red Team, declined to share specific details of the model’s cybersecurity successes, but said in Anthropic’s blog post that in recent weeks, the Mythos Preview has identified “thousands of high-severity vulnerabilities, including some in every major operating system and web browser.” Anthropic’s blog post doesn’t mention keeping humans in the loop for cybersecurity sweeps of the model; In fact, it highlights that the model identifies weaknesses and evolves[ed] Many related exploits – completely autonomously, without any human guidance.
The existence of the Cloud Mythos Preview was first reported in a data leak last month, which Anthropic attributes to human error. Dianne Penn, head of product management at Anthropic, said: The Verge Said in an interview that the company “is taking steps in terms of strengthening our processes… This has nothing to do with software vulnerabilities in any way.”
The Mythos preview will be privately available to the company’s Glasswing partners, which include JPMorgan Chase, Broadcom, Cisco, CrowdStrike, the Linux Foundation, and Palo Alto Networks, in addition to approximately 40 other organizations that maintain or build software infrastructure. For now, Anthropic will help subsidize the cost of its use. The company says it will commit up to $100 million in usage credits, as well as donate $4 million directly to the Linux Foundation and the Apache Software Foundation, Cheng said. In the longer term, as Anthropic and other AI companies face pressure to turn a profit, the program could evolve into a paid service that provides a new revenue stream — if it works well enough for companies to continue using it.
Despite its recent highly public confrontation with the Trump administration, Anthropic also said in the release that it is “in continuing discussions with U.S. government officials regarding the Cloud Mythos Preview and its offensive and defensive cyber capabilities.” When? The Verge Asked what this means, Penn confirmed that the company has “briefed senior officials in the US government about Mythos and what it can do,” and that the company is still “committed to working closely with all different levels of government.” Cheng said that although Anthropic is “engaging” with the government, he declined to say who the company had briefed.
<a href