WSJ Tech News Briefing TNB Tech Minute: Anthropic Previews New AI Model Designed to Combat AI Cyberthreats
16 snips
Apr 7, 2026 A preview of Anthropic’s new Mythos AI built to spot and patch hardware and software vulnerabilities. Discussion of why the tool will not be widely released due to its ability to exploit bugs. Coverage of Intel’s production struggles and its U.S. equity ties. Update on Elon Musk’s legal move to direct potential damages to OpenAI’s nonprofit arm.
AI Snips
Chapters
Transcript
Episode notes
Anthropic's Mythos Targets Infrastructure Bugs
- Anthropic built Mythos to find and patch hardware and software bugs across critical infrastructure.
- The preview goes to about 50 organizations including Amazon, Microsoft, Apple, Google, and the Linux Foundation, but won't be public due to its exploit capabilities.
Mythos Is Too Powerful For Public Release
- Mythos is designed to defend against malicious AIs that could scan for security vulnerabilities and cause widespread disruption.
- Anthropic's Frontier Red Team says Mythos is so capable at finding and exploiting bugs that the company won't release it to the public.
Intel Partners With Musk For TerraFab Chips
- Intel will partner with Elon Musk's companies to design and scale ultra-high chips for SpaceX XAI and Tesla under TerraFab.
- The deal helps Intel regain footing after cutting capacity while rivals like NVIDIA and AMD surged.