Anthropic Expands Project Glasswing to 150 New Organizations

Anthropic expanded Project Glasswing to include 150 new organizations, including critical infrastructure providers, after initial partners found over 10,000 security flaws.

Anthropic announced the expansion of Project Glasswing, increasing the number of participating organizations from approximately 50 to 150. The initiative focuses on enhancing cybersecurity by leveraging Claude Mythos Preview to identify and mitigate software vulnerabilities. The new partners, based in more than 15 countries, include entities that provide critical infrastructure and maintain widely used codebases. This expansion aims to bolster global cybersecurity as AI models like Mythos Preview become more prevalent. Source: anthropic

The expansion follows several weeks of collaboration with security experts, open-source maintainers, and the US government. Anthropic emphasized that the new partners represent industries not well-represented in the initial cohort, such as power, water, healthcare, communications, and hardware. Many of these organizations are vendors that support numerous other entities worldwide, including governments. The company highlighted that a successful attack on these partners' codebases could affect over 100 million people, with significant implications for global and national security. Source: anthropic

Project Glasswing aims to help the software industry adapt to the evolving cybersecurity landscape shaped by AI. Anthropic noted that the initiative has sparked conversations about how AI is transforming cybersecurity, influencing both the expansion of the program and its long-term goals. The company warned that without proper safeguards, AI models could enable more frequent and unpredictable cyberattacks. As part of its strategy, Anthropic plans to shift its support from vulnerability detection to patching and deploying secure software. Source: anthropic

Key points

Anthropic expanded Project Glasswing to include approximately 150 new organizations.
Initial Project Glasswing partners found more than 10,000 high- or critical-severity security flaws.
The new partners include entities that provide critical infrastructure and maintain widely used codebases.
A successful attack on these partners' codebases could affect over 100 million people.
Project Glasswing aims to help the software industry adapt to the evolving cybersecurity landscape shaped by AI.
Anthropic warned that without proper safeguards, AI models could enable more frequent and unpredictable cyberattacks.

WRITTEN BY

Marcus Feld

AI Business & Markets

Marcus reports on AI funding, acquisitions, partnerships, and the business behind the technology.