Human debut for ‘Project Glasswing’ and a new AI model for cybersecurity

Anthropic is launching 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 described as a way for large companies, and perhaps even the government, to identify vulnerabilities in their systems with almost no human intervention.

Anthropic is offering its launch partners access to the Claude Mythos Preview, a new general-purpose model that it does not currently plan to release publicly due to security concerns. Newton Cheng, cyber leader of Anthropic’s Red Frontier Team, said: Edge The model will give cyber defenders a “head start” against adversaries. Partners will use the model to analyze their system to detect high-risk vulnerabilities and help patch them. Access is restricted to prevent those same adversaries from using it to find vulnerabilities and carry out attacks.

Although Claude Mythos Preview was not specifically trained for cybersecurity purposes, Anthropic said in a statement that the model’s “strong programming and reasoning skills” are behind the advances in cybersecurity. In an interview with EdgeNewton Cheng, cyber lead for Anthropic’s Red Frontier Team, declined to share specific details about the model’s cybersecurity successes beyond the company. Publicly published examplesbut an Anthropic blog post said that in recent weeks, Mythos Preview has identified “thousands of high-risk vulnerabilities, including some found in every major operating system and web browser.” The Anthropic blog post makes no mention of keeping humans informed of the model’s cybersecurity campaigns; In fact, it highlights that the model identified the vulnerabilities “and developed (released) several related exploits – completely independently, without any human guidance.”

The existence of the Claude Mythos Preview was first reported last month In a data leak, which Anthropists attribute to human error. Diane Penn, Head of Product Management at Anthropic, said: Edge In an interview, the company “is taking steps in terms of enhancing our operations… and this has not been related to software vulnerabilities in any way.”

The Mythos Preview will be available exclusively to Glasswing’s corporate partners, which also include JPMorgan Chase, Broadcom, Cisco, CrowdStrike, Linux Foundation, and Palo Alto Networks, as well as about 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 $4 million in direct donations to the Linux Foundation and the Apache Software Foundation, Cheng said. In the long term, as Anthropic and other AI companies face pressure to turn a profit, the software could evolve into a paid service that provides a new revenue stream — if it works well enough for companies to continue using it.

Despite that The latest clash is very public With the Trump administration, Anthropic also said in the statement that it has “ongoing discussions with U.S. government officials about the Claude Mythos Preview and its offensive and defensive cyber capabilities.” when Edge When asked what that means, Ben confirmed that the company has “briefed senior officials in the US government about Mythos and what it can do,” and that the company remains “committed to working closely with all different levels of government.” Cheng said that although Anthropic “deals with” the government, he declined to speak to whom the company specifically briefed him.