tech
May 8, 2026
OpenAI Makes Its Rival to Anthropic's Mythos More Widely Available to Cyber Defenders
Recent security testing suggests that GPT-5.5 model is nearly as good at finding and exploiting software bugs as Anthropic's Mythos Preview.
TL;DR
- OpenAI is providing a more permissive version of GPT-5.5, codenamed 'Spud,' to vetted cyber defenders.
- Security testing shows GPT-5.5 is nearly as effective as Anthropic's Mythos Preview at finding and exploiting software bugs.
- The release has triggered urgent discussions on preventing the technology from falling into the hands of malicious actors.
- Vetted defenders will use GPT-5.5-Cyber to hunt for bugs, study malware, and reverse engineer attacks, with limitations on tasks like credential theft.
- The new model assists defenders in writing proofs of concept and running security posture simulations.
- Another version of GPT-5.5 is available to help with understanding code, mapping affected surfaces, and reviewing patches.
- Both OpenAI and Anthropic are developing approaches to manage the rollout of cyber-capable AI models.
- The White House is considering executive actions regarding federal involvement in future AI model rollouts.