Anthropic's Project Glasswing logged 10,000-plus confirmed critical vulnerabilities and expanded its partner list; Claude Code pushed a quiet infrastructure update.
Anthropic's Project Glasswing posted a milestone Friday: Claude Mythos Preview has confirmed more than 10,000 high-or-critical-severity vulnerabilities across production operating systems, browsers, and infrastructure software. (Benzinga, Anthropic) The program is expanding to additional partners beyond its founding roster: AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. Some partners reported bug-discovery rates more than 10 times their pre-Mythos baseline.
The number is notable. So is the framing: Anthropic is positioning this as a defender-first initiative, getting Mythos Preview's findings to the people who can patch them. That framing holds only as long as the output discipline does. On that question, the update is quiet.
Anthropic's AI-assisted vulnerability-discovery program reports more than 10,000 confirmed high-or-critical-severity findings across production software. The initiative is expanding to additional industry partners beyond the founding group, which includes AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. Some organizations report discovery rates more than 10x their baseline after integrating Mythos Preview into security testing workflows. Anthropic committed $100M in Mythos Preview usage credits and $4M in donations to open-source security organizations at the April launch; no new financial commitment was announced in this update. (Benzinga)
Internal infrastructure improvements only. No user-facing changes.
Daily digest at 9 PM ET. Weekly magazine every Friday morning. Six labs, one feed. No spam, one-click unsubscribe.