Something unusual has been happening around artificial intelligence over the last several weeks.
A remarkably consistent message is suddenly coming from researchers, major AI companies, politicians, advocacy organizations and the media: Artificial intelligence may be advancing too quickly, the consequences could be catastrophic, and governments need to intervene before it is too late.
Any one of those warnings could be taken at face value. Taken together, however, the timing and connections are worth examining.
I’ve been following this story closely, and what initially looked like a series of disconnected AI-safety stories has increasingly developed into a much larger policy battle.
In my VIP report at RedState, “Second Anthropic Researcher Abruptly Quits — His Next Move Is a Major Red Flag for AI,” I take a critical look at the departures of researchers Jacob Coxon and Joe Benton from Anthropic.
Both issued stark warnings about the potential dangers posed by increasingly capable AI systems. But I also examined reporting about the organizations and policy infrastructure surrounding those warnings, including Benton’s move to METR, an organization involved in evaluating the safety of advanced AI systems.
That raised a question that has followed this story ever since: What happens if today’s voluntary AI-safety organizations become tomorrow’s government-mandated gatekeepers?
Then came the politics.
