TLDR Tech / InfoSec
Anthropic grants Project Glasswing access to 150 more companies, with a focus on critical infrastructure
Anthropic expanded its Project Glasswing AI vulnerability-hunting initiative to 150 additional companies, concentrating on critical infrastructure across power, water, healthcare, communications, and hardware, where it estimates a successful codebase attack could affect more than 100 million people per partner. Analysts welcomed the wider net for bug discovery but warned that the real constraint sits downstream, since vendors and SOCs already struggle to validate, prioritize, and patch known issues, so a 10x or 100x jump in findings risks turning cybersecurity from a visibility problem into an execution one. Researchers also flagged trust gaps around the undefined "security requirements" that gate new participants and around confidence scoring for AI-generated patches, while one analyst cautioned that broadening access beyond a small, vetted group heightens leak exposure for a model that has already had two reported leaks.
Read the full story