Import AI 459: AI oversight is difficult; scaling laws for protein folding models; and pricing the extinction risk of AI systems

# AI Oversight Remains Unsolved as Systems Grow More Powerful Controlling what AI systems do is turning out to be much harder than expected, especially as these tools become more capable and widely deployed. At the same time, researchers are making progress on specialized AI for tasks like predicting protein structures, though scaling these systems comes with its own challenges. The combination of these advances—more powerful AI with less certainty about how to oversee it—has sparked serious conversations about what risks we might be taking on.