Amid growing concerns about artificial intelligence creating new cybersecurity vulnerabilities, Anthropic has announced a new initiative called Project Glasswing aimed at using AI defensively to protect critical software systems.
The project, which counts Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks as partners, seeks “to secure the world’s most critical software” against AI-powered attacks, according to the company.
Participants in Project Glasswing will use Claude Mythos Preview, an unreleased general-purpose model from Anthropic, to enhance their own security projects. Anthropic claims the model has already identified thousands of exploitable vulnerabilities, “including in every major operating system and web browser.”
The company stated its intention to begin using its tools defensively to prevent malicious AI applications that could have severe consequences for economies and security.
Anthropic has been pretty vocal about AI ethics compared to other companies. Earlier this year, it even turned down a Pentagon request to loosen its guardrails, which reportedly led the Department of Defense to slap it with a “supply chain risk” label.
While Project Glasswing marks a move toward stronger cybersecurity in the AI era, some incidents have already raised concerns. In February, Anthropic’s Claude model was reportedly used by a hacker in attacks targeting multiple Mexican government agencies.
Maybe you would like other interesting articles?