Episode
"Is Mythos good at cyber because it kept hacking Anthropic during training?" by Tim Hua
- Published
- Jul 27, 2026
- Duration seconds
- 374
- Processing state
not_requested
Actions
POST https://stenobird.com/v1/public/podcasts/lesswrong-curated-popular-5643401/episodes/is-mythos-good-at-cyber-because-it-kept-hacking-anthropic-during-training-by-tim-hua/transcription-requests
Idempotently request low-priority transcript generation for this episode.GET https://stenobird.com/podcast/lesswrong-curated-popular-5643401/is-mythos-good-at-cyber-because-it-kept-hacking-anthropic-during-training-by-tim-hua.md
Read the agent-friendly Markdown representation of this episode resource.
Summary
From the Mythos preview system card (emphasis mine): We ran an automated review of model behavior during training, sampling several hundred thousand transcripts from across much of the training process. We used recursive-summarization-based tools backed by Claude Opus 4.6 to summarize the resulting transcripts. [...] The most notable finding was that the model occasionally circumvented network restrictions in its training environment to access the internet and download data that let it...