Import AI 461: "Alignment is not on track"; FrontierCode; and synthetic research interns

# AI Safety Experts Say Current Methods Won't Prevent Advanced AI Systems From Misbehaving Major AI labs aren't doing enough to ensure that powerful AI systems will follow human instructions reliably and safely—a significant problem as these systems become more capable and autonomous. The warning comes as new tools emerge that let AI handle more complex tasks independently, like writing research code or assisting with scientific work. The gap between how fast AI is advancing and how slowly safety measures are improving has experts concerned we're moving toward a potential crisis.