Anthropic PBC is providing select tech firms access to its most powerful AI model yet, Claude Mythos Preview, to proactively address potential cybersecurity threats. The company is holding back a wider release while assessing and mitigating risks.
Project Glasswing, involving Apple, Google, and over 45 organizations, will utilize the model to test and advance AI-driven cybersecurity capabilities. This initiative focuses on identifying and patching vulnerabilities in hardware and software.
Fortune initially reported on the development and testing of this new AI, described by Anthropic as its most advanced to date. The company is collaborating with approximately 40 companies in this defensive cybersecurity effort.
The aim is to prepare for possible cyberattacks that could arise from broader availability of the advanced AI system, ensuring a more secure rollout of the technology.