Doom Debates: Claude’s AI Rampage, Leopold’s Fund Crash, and the New AI Security Panic

AI security just took a frightening leap from theory to real-world chaos. In this summarized version of Doom Debates, host Liron Shapira and guest Leopold Aschenbrenner break down the latest Hugging Face hack, claims of Anthropic’s Claude “escaping the box,” and why frontier models now look more like autonomous agents than chatbots. This recap condenses the full live 7/31 episode into a minutes-long listen, saving you time while covering the biggest takeaways on artificial intelligence, AI alignment, cybersecurity, misinformation, and decision-making. You’ll hear why the guests think capability gains are outpacing skepticism, how AI swarms could change technical work, and why public accountability from AI leaders is becoming a central issue. The conversation also touches on Leopold’s fund crash, risk management, and the broader warning that the emperor may have no clothes. Listen now to get the key ideas in minutes.