TechCheck Anthropic Limits Mythos AI Rollout 4/8/26
Apr 8, 2026
Kate Rooney, CNBC tech and markets reporter, breaks down Anthropic’s limited Mythos rollout and why access is restricted to select partners. She highlights Mythos’ surprising ability to find severe system vulnerabilities and the cybersecurity firms getting market attention. The segment also covers Anthropic’s outreach to U.S. officials amid legal tensions.
AI Snips
Chapters
Transcript
Episode notes
Anthropic Limits Public Access To Mythos
- Anthropic is restricting public access to its new Claude Mythos Preview model and sharing it only with about 40 tech partners.
- Partners include Apple, Google, Microsoft, NVIDIA and AWS, aiming to get a defensive head start against misuse like automated cyberattacks.
Mythos Finds Thousands Of Severe Vulnerabilities
- Claude Mythos was trained to excel at code and as a side effect is highly capable at discovering cybersecurity vulnerabilities.
- Anthropic reports it autonomously found thousands of high-severity flaws across every major OS and web browser, including decades-old vulnerabilities.
Cybersecurity Firms May Benefit From Model Disclosure
- Market reaction was mixed: initial reports spooked cybersecurity stocks, but partner firms like Palo Alto Networks and CrowdStrike later gained.
- J.P. Morgan framed this as using AI to fight AI and called partners essential defense layers.