CEOs held Mythos call
Executives from Google, OpenAI, Microsoft, CrowdStrike and U.S. government officials held a call after concerns surfaced about Anthropic’s unreleased model ‘Mythos’ and its potential to expose cybersecurity vulnerabilities. The Times of India report said the discussion contributed to restricting the model’s release while the risks were assessed. (timesofindia.indiatimes.com)
Top United States officials held a private call with the chiefs of Google, OpenAI, Microsoft, CrowdStrike and others before Anthropic rolled out its gated Mythos model. (cnbc.com) The April 10 CNBC report said Vice President JD Vance and Treasury Secretary Scott Bessent joined the call, and participants included Anthropic chief executive Dario Amodei, OpenAI chief executive Sam Altman, Google chief executive Sundar Pichai, Microsoft chief executive Satya Nadella, CrowdStrike chief executive George Kurtz and Palo Alto Networks chief executive Nikesh Arora. Reuters matched that account the same day. (cnbc.com) (usnews.com) The discussion centered on two questions: how secure large language models are, and what governments and companies should do if artificial intelligence starts giving attackers an edge over defenders in cyberattacks. Anthropic had already decided not to release Mythos broadly. (cnbc.com 1) (cnbc.com 2) A software vulnerability is a hidden flaw in code, like a weak lock in a building door. Anthropic said Mythos is unusually good at finding and exploiting those flaws, and said the model has already found thousands of high-severity vulnerabilities, including bugs in major operating systems and web browsers. (anthropic.com) Instead of a public launch, Anthropic put Mythos into Project Glasswing, a restricted program announced on April 7 with launch partners including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, Microsoft, Nvidia and Palo Alto Networks. Anthropic said it also extended access to more than 40 additional organizations that build or maintain critical software infrastructure. (anthropic.com 1) (anthropic.com 2) Anthropic said it is committing up to $100 million in usage credits for the project and $4 million in direct donations to open-source security groups. Dianne Penn, Anthropic’s head of research product management, told CNBC there was “a lot of internal deliberation” before the company limited access. (anthropic.com) (cnbc.com) The White House call was not the only government meeting tied to Mythos. Reuters reported on April 9 that Bessent and Federal Reserve Chair Jerome Powell convened bank chiefs to warn about cyber risks linked to Anthropic’s latest model. (wtaq.com) The backdrop is a wider fight over how far frontier artificial intelligence labs should go in releasing systems that can write code, probe networks and automate security research. Anthropic said the work of defending critical software “might take years,” while model capabilities are likely to advance “over just the next few months.” (anthropic.com) Anthropic has not framed Mythos as a consumer product. For now, the model is being treated more like sensitive security equipment: shared with a small circle, watched closely by Washington, and kept off the open market while the risks are tested. (cnbc.com 1) (cnbc.com 2)
