Anthropic Wants to Slow Down AI: Real Safety or Regulatory Capture?

In this episode, I analyze Anthropic’s most controversial shift: while Dario Amodei warns that AI is advancing at an exponential rate, he is also proposing that the government have the authority to block or reverse the deployment of frontier models that do not pass independent safety tests.

I review the essay “Policy on the AI Exponential,” the four risk categories Anthropic highlights — biological, cyber, loss of control, and automation of AI R&D itself — and the underlying argument: if this technology can cause severe harm, it should not enter the market without external auditing and a genuine ability to stop it.

But I also examine the harshest criticism: the possibility of regulatory capture, in which safety rules ultimately favor large companies that can afford evaluations, compliance, and advanced model protection.

I also connect this debate to Amodei’s broader position in “The Adolescence of Technology,” where he argues that humanity is entering a turbulent period that will test its ability to handle unprecedented technical power.