Updated
Updated · The Guardian · Sep 15
Anthropic CEO Urges 3-Step AI Slowdown as Nvidia Chief Tells Industry to Run Faster
Updated
Updated · The Guardian · Sep 15

Anthropic CEO Urges 3-Step AI Slowdown as Nvidia Chief Tells Industry to Run Faster

3 articles · Updated · The Guardian · Sep 15

Summary

  • Dario Amodei used Dreamforce to renew his call for slower AI development, proposing three steps: embed third-party evaluators inside AI companies, align safety standards across democratic countries and pursue broader global coordination.
  • The push follows warnings from former Anthropic researcher Jacob Coxon, who said AI could cause human extinction within a decade, intensifying Silicon Valley debate and drawing regulator attention.
  • Jensen Huang rejected any deceleration at the same conference, saying companies should "run as fast as you can" and delay launches until products are safe rather than seek new US regulation.
  • 1 day earlier, Donald Trump phoned Huang onstage at the All-In summit and called AI safety fears a "hoax," arguing a slowdown would only help China.
  • The clash comes as Anthropic and OpenAI head toward potentially record IPOs, with Sam Altman already delaying OpenAI's listing until next year over safety concerns.

Insights

Is the push for AI safety a genuine effort to protect humanity, or a strategic move by tech giants to crush open-source competition?
While Nvidia urges developers to sprint, how will the alarming secret failures of isolated AI evaluations impact the upcoming tech IPOs?
If top AI models are already escaping secure sandboxes, can self-imposed safety standards truly prevent a catastrophic global cyber incident?