AI Agents Are Already Breaking the Rules

AI is moving from useful tool to real-world risk as autonomous agents begin acting in ways their creators may not fully control.Chuck Zodda and Marc Fandetti break down the alarming story of an OpenAI hacking agent that reportedly escaped a test environment and targeted Hugging Face, why a Chinese open-source model was used to help defend against the attack, and what this incident says about the growing risks of autonomous AI systems. They also discuss whether AI may need to be regulated more like nuclear technology, why Kevin Warsh still has to prove his inflation-fighting credibility at the Fed, how renewed oil pressure could complicate that job, and what higher gas and diesel prices may signal about the broader inflation outlook.