Anthropic isn’t sure if it will sell its powerful new Mythos model to the public, the AI startup’s co-founder said at Semafor World Economy, days after top US officials summoned Wall Street banks to raise security concerns about the tech. Mythos sparked alarm at the company when it discovered previously unknown security flaws in every major web browser and operating system, unnerving global financial authorities. Mythos is “not a special model,” Jack Clark said Monday; within a year and a half, “there will be open-source models from China that have these capabilities.” Anthropic is working with the US government to address potential cyber risks from Mythos despite an ongoing legal battle with the Pentagon, which Clark dismissed as a “narrow contracting dispute.”

Semafor

https://www.semafor.com/newsletter/04/13/2026/semafor-flagship-the-next-test?utm_source=newslettershare&utm_medium=flagship+asia&utm_campaign=flagshipnumbered4#f