Anthropic Expands AI Cybersecurity Program to 150 More Organizations
Power, healthcare, water, communications and hardware providers are among the new participants in Anthropic’s Project Glasswing.
Topics
News
- SAP Labs India Selects 16 Startups for 2026 Startup Studio Cohort
- Nearly Half of Workers Now Spend More Time Managing AI Than Doing Tasks, Study Finds
- Microsoft Shows AI Wearable Badge, Desktop Device Concepts
- Anthropic Expands AI Cybersecurity Program to 150 More Organizations
- Alphabet Plans $80 Billion Equity Raise to Fund AI Infrastructure
- Anthropic Files Confidentially for IPO
Anthropic said Tuesday it is expanding Project Glasswing, its AI cybersecurity initiative, to about 150 more organizations across more than 15 countries, extending access to a powerful model designed to find and help fix software vulnerabilities.
The expansion will take Project Glasswing from about 50 partners to about 200. Each new participant should meet Anthropic’s security before receiving access to Claude Mythos Preview, the company’s advanced cybersecurity-focused AI model.
The move comes a day after Anthropic had confidentially filed for an initial public offering after a recent funding round valued the AI company at about $965 billion.
Anthropic said the first group of Project Glasswing partners, which received access in April, has used Mythos Preview to identify more than 10,000 high- or critical-severity security flaws.
The new cohort includes organizations in power, water, healthcare, communications and hardware, sectors that Anthropic said were not well represented in the program’s first phase.
Many of the participants maintain software and infrastructure used by governments, companies and other institutions worldwide.
“What each partner has in common is that a successful attack on their codebase could be catastrophic,” Anthropic said. “For most partners, we estimate that a major attack could affect more than 100 million people, with important ramifications for both global and national security.”
Anthropic said Project Glasswing is meant to give defenders time to adapt before similarly powerful AI systems become more widely available. The company said it expects other AI companies to have Mythos-class models within six to 12 months and warned that some could be released without safeguards to prevent misuse.
The company said partners are already using Mythos Preview not only to find vulnerabilities, but also to write patches, run pre-release checks, conduct penetration testing, automate threat detection and help rebuild legacy codebases in memory-safe languages.
Anthropic recently launched Claude Security, a product that uses its latest public models to scan code and suggest fixes. It is also making specialized tools available on request to trusted security teams.
The company said it plans to keep expanding Project Glasswing to more critical infrastructure operators, open-source software maintainers and cybersecurity researchers.
“We intend for future expansions to cover organizations in the US and overseas, just as this one does,” the company said. “If we’re successful, we hope to enable a permanent advantage for defenders.”
It is also working to broaden its Cyber Verification Program, which is intended to give more defensive security teams access to Mythos-level capabilities.
