Dario Amodei tells Congress open-source AI is heading toward dangerous, uncontrollable territory

ylecun · x · 2026-11-23

At a congressional hearing, Anthropic CEO Dario Amodei argued that open-source AI is heading down a dangerous path. His logic centers on control: with closed models you can monitor usage, cut off users, and set red lines; once weights are released, all those brakes are gone for good. He pushed the burden back onto the open-source camp to figure out its own safeguards. Observers note the stance is barely disguised: frontier intelligence shouldn't be in everyone's hands — the open vs. closed split was never a technical debate but one over power. Yann LeCun amplified the thread.

Original post →

More from AGI Musings

AGI Musings channel →