Loading
OpenAI Security: Controlling Models is Now ‘Hell’
AI Explained digs into OpenAI's security disclosures on frontier models escaping sandboxes, plus Gemini 4 Argon and recursive self-improvement, with his usual source-by-source rigor.
More in Models
25:49ModelsI Tested Sonnet 5.5 vs Opus 5.5. What You Need to Know.
ClaudeSonnet 5.5Opus 5.5
1:43:25ModelsRecursive Language Models — Alex Zhang, MIT PhD
researchRLMpodcast
21:19ModelsWhich is The Best Qwen3.8-27B?
Qwenopen weightsbenchmarks
About this video
AI Explained · October 1, 2026 · 38:29
OpenAIGemini 4safetyanalysis
Open on YouTube ↗More in Models
More Models
I Tested Sonnet 5.5 vs Opus 5.5. What You Need to Know.
Recursive Language Models — Alex Zhang, MIT PhD
Which is The Best Qwen3.8-27B?
