Anthropic expands AI security program to shield critical infrastructure

Anthropic on Tuesday announced the expansion of its Project Glasswing program, an initiative that provides artificial intelligence tools to organizations tasked with securing critical software systems worldwide. The company said it is adding approximately 150 new partners across more than 15 countries to the effort, with each organization required to meet strict security requirements before gaining access to specialized AI systems.

The AI firm noted that the program launched in early April with roughly 50 partners who have since detected more than 10,000 high- or critical-severity flaws in their codebases using Claude Mythos Preview. The specialized model scans software systems to identify vulnerabilities before they can be exploited, with partners drawn from sectors including power, water and health care.

Critical infrastructure protection

The new participants include organizations providing critical infrastructure across sectors such as power, water, health care and communications, many of which maintain codebases relied upon by governments and other institutions. Anthropic said that for most partners, a successful cyberattack on their codebase could affect more than 100 million people, carrying implications for national and global security.

The expansion follows extensive collaboration with existing Project Glasswing partners, the broader security industry, open-source software maintainers and the US government, according to the company. Anthropic told Anadolu Agency that many of the new participants are vendors, companies or nonprofits that maintain codebases relied upon by other organizations, including government agencies.

Emerging AI threats and tools

Anthropic warned that "cheap, fast AI models with powerful cyber capabilities are around the corner," noting that similar systems could become widely available within six to 12 months, including to malicious actors. The company said it has launched Claude Security — a product that uses publicly available models to scan codebases and suggest patches — to help organizations adapt to these emerging threats.

Anthropic stated that it plans to continue expanding Project Glasswing geographically while scaling up related cybersecurity initiatives. "If we're successful, we hope to enable a permanent advantage for defenders," the company said, adding that future efforts will focus on improving vulnerability detection, disclosure and patching processes across the industry.