It's time to panic about AI safety

“Imagine only one person had a superintelligent lawyer. He would have an unfair advantage in court, even if his position was wrong on the merits.” That thought experiment, offered by Mark Zuckerberg in a Wall Street Journal op-ed, was meant to argue for open-weight AI and optimism about the technology. But on this episode of The Vergecast, David Pierce and Nilay Patel use it as a launching pad for a much darker conversation. They argue that the AI industry is entering a dangerous phase of competitive panic, where companies like OpenAI, Meta, and a wave of cheaper Chinese model makers are all racing ahead without meaningful guardrails—and that the U.S. government, broken and distracted, is incapable of providing them. The episode weaves together three major threads: the hardware and gadget consequences of the AI boom, the philosophical blind spots of tech leaders, and the very real safety failures that are starting to pile up.

The gadget squeeze: why everything is getting more expensive