Anthropic plans to make Mythos-class AI models widely available
Anthropic plans to expand access to its Mythos-class AI models, which have already uncovered decades-old vulnerabilities in critical software like OpenBSD and FreeBSD, despite White House opposition. The company argues controlled access through Project Glasswing—currently limited to partners like AWS, Apple, and Microsoft—is safer than restricting the technology entirely.
Anthropic is advancing plans to make its Mythos-class AI models more widely available, despite concerns from the White House. The company’s Claude Mythos Preview, announced on April 7, 2026, can autonomously identify and exploit zero-day vulnerabilities in operating systems, web browsers, and open-source projects. It has already uncovered thousands of vulnerabilities, including a 27-year-old flaw in OpenBSD and a 17-year-old remote code execution issue in FreeBSD. Currently, access is restricted through Project Glasswing, a controlled initiative involving AWS, Apple, Microsoft, Google, NVIDIA, CrowdStrike, and the Linux Foundation. Anthropic has committed up to $100 million in usage credits to support the project’s defensive applications. The White House opposed expanding access to roughly 70 additional organizations on April 30, 2026, citing risks of misuse. Anthropic counters that broader access—under safeguards—could accelerate vulnerability patching before malicious actors exploit them. Project Glasswing partners, including CrowdStrike and the Linux Foundation, could integrate Mythos capabilities into cybersecurity products, potentially strengthening their competitive position. The Linux Foundation’s involvement highlights the urgent need to address vulnerabilities in open-source projects, which often lack sufficient resources. The $100 million investment in Glasswing serves as a test for responsible deployment. If the initiative demonstrates measurable improvements in vulnerability detection and patching, it could bolster Anthropic’s case for wider release. However, the White House’s stance suggests political approval remains uncertain despite technical safeguards.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.