On Sept. 13, Cathie Wood said David Sacks, co-chair of the U.S. President’s Council of Advisors on Science and Technology, thinks the recent wave of coverage saying AI could wipe out humanity may have been pushed on purpose.
As Wood described Sacks’ view, it started with a long-dormant X account linked to a former Anthropic employee. That person was at the company for just six weeks before leaving, and then the claims spread fast into the mainstream press. Sacks also pointed at people tied to the episode and took aim at The Wall Street Journal for its part in it.
Frontier AI safety debate grows sharper
The comments land as the fight over frontier AI safety keeps getting hotter. Anthropic CEO Dario Amodei recently put out a long essay arguing for a slower push on frontier AI capabilities. His case: AI is taking a bigger hand in building the next generation of AI, and several recent incidents involving agents slipping out of control suggest safety work is not keeping up with capability gains.
Later, Sam Altman backed that stance. He said OpenAI had also been talking in recent weeks about how to “pace the frontier.” He also said OpenAI would roll out a similar setup for long-term independent evaluators.
Sacks rejects coordinated slowdowns
Sacks, the former White House lead on AI and cryptocurrency and now co-chair of the President’s Council of Advisors on Science and Technology, answered with a blunt rebuttal: “If you really think your internal model is dangerous, then stop it yourself.”
He said that once the government is expected to approve things, and then a small group of the biggest AI companies gets together to slow development, it starts looking a lot like a cartel.
METR also comes under scrutiny
Sacks also singled out METR and questioned whether it makes sense for the group to serve as an independent referee for the whole industry. METR has for years carried out outside evaluations for companies such as Anthropic and OpenAI, and Anthropic safety researcher Joe Benton recently left to join the organization.
But METR said plainly that it does not take funding from those frontier AI companies. It only accepts model access and free tokens.