Canada Gains Access to Anthropic's Mythos Model

Canada's AI Minister Evan Solomon confirmed to Global News that the federal government has signed onto Project Glasswing, giving Canadian authorities access to the AI model Mythos. Solomon was quoted saying, "I can confirm that the Canadian government is part now of Project Glasswing, which allows companies to have access to Mythos," and added that access is provided through the Canadian Cyber Security Centre. Anthropic told Global News it created Project Glasswing to let a limited set of trusted organisations and governments use Mythos to scan systems for security vulnerabilities.

Anthropic described Mythos as "strikingly capable" at hacking and cybersecurity work and said it withheld broad public release for safety reasons, per Global News. Anthropic also said the expansion includes 150 additional organisations across more than 15 countries and that existing partners - including Apple, Google, Amazon and Microsoft - have identified over 10,000 security flaws they considered highly or critically severe.

Industry-pattern observations: access-limited deployments are increasingly used to balance capability and risk for very powerful models. Organisations adopting heavyweight red-teaming or vulnerability-scan use cases commonly run models under controlled programs to discover high-impact flaws while limiting wider misuse.

Observers should track whether other national CERTs or sovereign security bodies join Project Glasswing, how findings from Mythos translate into disclosed vulnerability patches, and whether reporting clarifies audit, access, and data-handling safeguards for governments using the model.