Anthropic model flags OS and browser vulnerabilities
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic announced Project Glasswing, a cybersecurity partnership with Nvidia, Google, Amazon Web Services, Apple, Microsoft and other firms to let large companies and governments flag system vulnerabilities with virtually no human intervention.
- Claude Mythos Preview — Anthropic’s new general‑purpose model — autonomously flagged “thousands of high‑severity vulnerabilities” and even generated related exploits, including flaws in every major operating system and web browser, according to Anthropic’s blog.
- Anthropic is restricting access to Project Glasswing and Claude Mythos Preview to “defensive security” partners such as JPMorgan Chase, Broadcom, Cisco, CrowdStrike, the Linux Foundation, Palo Alto Networks and about 40 other software‑infrastructure organizations.
- Anthropic will subsidize partner usage with up to $100 million in usage credits and will donate $4 million to the Linux Foundation and the Apache Software Foundation.
- Anthropic has briefed senior U.S. government officials about Claude Mythos Preview’s offensive and defensive cyber capabilities and says it remains committed to working closely with government at all levels.
- Anthropic said the earlier data leak of Mythos Preview was caused by human error and that the company is taking steps to solidify its processes, though the leak was unrelated to software vulnerabilities.
Why it matters: Defensive security teams at firms like JPMorgan Chase and open‑source foundations now receive an AI that can autonomously discover and even craft exploits for high‑severity OS and browser bugs, giving them a rapid patching advantage while Anthropic secures a $100 million usage‑credit commitment and strengthens its foothold in enterprise cyber‑defense.
