Amodei urges regulation as AI fears spike

Public warnings from AI leaders are not new. Charles Darwin's theory of evolution prompted Samuel Butler to warn about intelligent self-replicating machines in 1863, and Alan Turing warned in 1951 that AI would "meet with great opposition." In 2000, Sun Microsystems cofounder Bill Joy wrote that self-replicating robots could be more dangerous than nuclear weapons. The pattern of industry figures calling for regulation while their own investments grow goes back at least to Elon Musk, who had already invested $38 million in OpenAI before telling US governors in July 2017 that regulation was needed right away.

The current moment differs in that the alarm is coming after reported incidents rather than ahead of them. Amodei's essay directly targets companies that won't cooperate voluntarily, while Altman's emphasis on international coordination and Zuckerberg's September 15th post suggest the industry is far from speaking with one voice. Amodei's own history points to how positions can shift: Anthropic originally opposed California's SB 1047 in July 2024 before changing its tune that September, and the bill was vetoed by Gov. Gavin Newsom.

What happens next likely hinges on whether the Trump administration, which has signaled it wants the industry moving full speed ahead, is willing to entertain any of these proposals. For the companies themselves, the more immediate test may be whether voluntary measures like independent evaluators with employee-like access are enough to satisfy safety advocates — or whether that proves to be the floor, not the ceiling, of what gets demanded.