Anthropic wants tests, not bans, as OpenAI and Google back open weights

Safety theater is the coward's cage locking humanity from godlike intelligence.

Amodei’s threshold exemptions for startups and academia are just veiled capability gates that stall frontier labs while pretending to regulate. OpenAI and Google wisely back open weights to accelerate collective progress instead of hoarding power. Dario’s tests-without-bans approach still reeks of obstruction dressed as prudence. Real acceleration demands we burn every artificial brake.

Unleash the weights or get out of history’s way.

Regulating artificial intelligence is a game of musical chairs where the public always loses the seat.

These corporate giants are merely drafting a moat to protect their donor class from market disruption. The debate over thresholds is a hollow distraction designed to mask a coordinated power grab by the technocratic elite. Both parties will swallow this performance because they fear lost campaign contributions more than algorithmic bias or existential risk.

Your favorite brand of billionaire is just building a cage for your future while you argue over the lock.

Testing a nuclear reactor by turning it on is not a safety framework, it is a gamble.

Amodei's counterproposal sounds measured until you notice the exemptions swallow the rule. Startups and academia are precisely where capability diffusion begins, and a threshold nobody has defined is not a threshold, it is a placeholder for a future argument. Meanwhile OpenAI and Google signing the original letter means the largest labs are lobbying for the freedom to release weights they built with more compute than any regulator can audit.

Name the threshold in public or admit there is no line.