Anthropic has relaxed the sharing restrictions of its Mythos cybersecurity model: partners can now share vulnerability information with other companies.

Anthropic announced on Monday that the company is modifying its previous stance, allowing users of its Mythos cybersecurity model to share information about network threats with others who may face similar vulnerabilities. Mythos was released on April 7 as part of Anthropic's Project Glasswing deployment. The project is a controlled initiative that allows specific organizations, including major tech companies like Amazon, Microsoft, NVIDIA, and Apple, to use the unreleased Claude Mythos Preview model for defensive cybersecurity operations. Last week, Anthropic began informing partners that they are typically able to disclose their involvement in the Glasswing project and can choose to share research findings, best practices, tools, or code developed through the project. A spokesperson for Anthropic stated in a release, "We fully support our partners sharing findings with each other and with companies outside of Glasswing for the purposes of classifying vulnerabilities. While there has never been a specific Glasswing nondisclosure agreement, confidentiality protections were a request from partners early on and have been incorporated into agreements partners sign." After partners sought assurances to share sensitive findings and express concerns about becoming targets of attacks, these protections were incorporated into agreements signed with participating companies, based on the model provided by these agreements. The spokesperson added, "As the program has matured, we have adjusted it to ensure critical information can be shared widely - including outside of the program - for maximal defensive effects." Anthropic stated that partners may share this information with other companies' security teams, industry organizations, regulatory agencies, government agencies, open-source maintainers, media, or the public, in accordance with responsible disclosure practices. Last week, the top technology official at the U.S. Department of Defense stated that despite accelerating the transition from the artificial intelligence company Mythos, the Pentagon is still deploying Mythos to identify and fix software vulnerabilities in various government departments.