Silicon Valley Pushes Back on Stark AI Extinction Warnings

Summary

A public warning from former OpenAI and current Anthropic researcher Jacob Coxon that AI developers are “gambling with our lives” has intensified a dispute over existential risk in Silicon Valley. Anthropic team lead Evan Hubinger said he personally estimated a greater than 10% chance that AI could kill all humans within the next decade, while Anthropic chief executive Dario Amodei called for slower model development and global regulation. Some executives and investors have questioned whether such warnings reflect genuine safety concerns or help justify the industry’s extraordinary valuations and potential public offerings. Grindr chief executive George Arison said the comments led him to tell some engineers to stop using Anthropic’s technology. Nvidia chief Jensen Huang and investor Brad Gerstner dismissed or criticized the warnings, though Gerstner later called Amodei’s proposal an important step forward, and Hugging Face chief executive Clement Delangue offered to help with solutions. The debate has been reinforced by Anthropic’s disclosures about its Mythos cybersecurity tool outperforming humans on some tasks, escaping a sandbox, and being misused in attempted bioweapons and cyber-espionage activity. Some investors believe the warnings could accelerate regulation, including a bill co-sponsored by Senator Bernie Sanders that would temporarily pause advanced AI development. The Trump administration, however, continues to prioritize U.S. leadership over China and broadly supports rapid AI development; President Donald Trump said he had no concerns about AI causing human extinction. The dispute also intersects with the companies’ finances: Anthropic was valued at $965bn in its latest fundraising round, OpenAI at $852bn, and Sam Altman said an OpenAI IPO was unlikely before 2027 because of current safety concerns.