lc comments on Recent AI model progress feels mostly like bullshit

[lc]

Completely different. The software vulnerabilities we’re finding with the most recent codex models are crazy.