Apple joins Project Glasswing

Apple has joined Project Glasswing, a new Anthropic-led effort that gives major tech companies early access to an advanced artificial intelligence security model. (anthropic.com) Anthropic announced Project Glasswing on April 7, 2026, with Apple, Amazon Web Services, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks as launch partners. Anthropic said the group is focused on securing “the world’s most critical software.” (anthropic.com) The project uses Claude Mythos Preview, an unreleased model Anthropic said can find and exploit software flaws at a level beyond most human experts. Anthropic said the model has already found thousands of high-severity vulnerabilities, including flaws in every major operating system and web browser. (anthropic.com) Software vulnerabilities are hidden mistakes in code, like weak locks inside systems that run banks, hospitals, power grids, and web services. Anthropic said Glasswing is meant to help defenders find those weak points before attackers use similar tools against them. (anthropic.com) Anthropic said it is extending access beyond the launch partners to more than 40 additional organizations that build or maintain critical software infrastructure. The company also committed up to $100 million in usage credits and $4 million in direct donations to open-source security groups. (anthropic.com) CyberScoop reported that Anthropic’s early testing found a 27-year-old bug in OpenBSD and a 16-year-old flaw in FFmpeg, a widely used video software project. The report said Anthropic had contacted maintainers and that the disclosed vulnerabilities had been patched. (cyberscoop.com) The Linux Foundation said part of the money will support open-source projects that small maintainer teams often struggle to audit on their own. Its April 7 post said Project Glasswing gives maintainers access to a frontier model for defensive testing on both first-party and shared open-source code. (linuxfoundation.org) Anthropic said it does not plan to release Mythos Preview to the general public. CNBC reported the company limited rollout because of fears that a model with strong autonomous hacking ability could also be misused by attackers. (cnbc.com) Apple has not published a separate statement detailing which of its products or internal systems it will test through Glasswing. For now, the company’s role is as one of the named launch partners using the model in defensive security work alongside its own tools. (anthropic.com)