Anthropic launches Project Glasswing, an effort to prevent AI cyberattacks with AI (www.engadget.com)

Anthropic has launched Project Glasswing, a significant initiative aimed at fortifying critical software against AI-driven cyberattacks. Collaborating with tech giants such as Amazon Web Services, Google, and Microsoft, the project leverages Anthropic's unreleased Claude Mythos Preview model. This model reportedly identifies thousands of security vulnerabilities across major operating systems and web browsers, positioning it as a crucial tool in the fight against the rising threat of AI-enabled cybercrime. The initiative emphasizes a proactive approach to cybersecurity, aiming to mitigate potential risks that AI poses to economies and national security. This move is particularly notable in a landscape where ethical concerns surrounding AI usage are increasingly pressing. Earlier this year, Anthropic faced backlash from the Pentagon for its refusal to lift restrictions on its technology, leading to a designation as a "supply chain risk." Despite these challenges, Project Glasswing represents a constructive step toward addressing AI's dual-edged sword of innovation and threat. The unresolved issues from previous AI misuse, such as the reported exploitation of Anthropic’s Claude model in attacks against Mexican government agencies, underscore the urgency of this initiative within the AI/ML community.