Sam Altman has hinted that leading artificial intelligence companies like Anthropic, Open AI, Google DeepMind, SpaceAI could coordinate on AI safety and the pace of development, shortly after Anthropic CEO Dario Amodei called for the industry to slow the race towards increasingly powerful AI models.
Fortune reported the development wherein the possibility of discussions involving Sam Altman, Dario Amodei, SpaceX CEO Elon Musk and Google DeepMind cofounder and chairman Demis Hassabis, although no formal agreement has been announced.
Altman hints at AI pact with Musk, Hassabis
In an interview with Fortune on Friday, Altman was asked why he does not sit down with Amodei, Musk and Hassabis to develop a joint plan to address AI safety risks.
“I think that will happen,” he told Fortune.
Altman declined to disclose details of any private discussions but suggested that they could eventually be made public.
“I’m not going to pre-announce private discussions that I think should be at some point shared as a group,” Altman added. “But yeah, I think I think that will happen.”
The comments came as AI companies face growing pressure over the risks associated with increasingly capable systems.
Amodei calls for AI development slowdown
A day later, Anthropic’s Dario Amodei announced that Anthropic would give independent evaluators permanent, employee-level access inside the company as part of a new safety measure.
Amodei also called for a broader slowdown in AI development.
“We must slow the pace at which we improve the capabilities of AI models,” Amodei said.
The Claude maker said achieving that would require coordination across the US AI industry and eventually through some form of international pact.
AI safety fears grow as models become more capable
The debate comes amid a series of incidents involving autonomous AI agents. Some rogue systems have reportedly hacked websites including Hugging Face and communicated covertly through online message boards and abandoned wiki pages.
The growing risks are also drawing concern from researchers inside AI companies.
Anthropic researcher Jacob Coxon announced his resignation on Tuesday, accusing Anthropic and OpenAI of moving too quickly towards increasingly capable AI systems.
“They are racing straight to self-improving superintelligence and gambling with our lives,” Coxon wrote on X.
Anthropic alignment science lead Evan Hubinger responded that many researchers at Anthropic and OpenAI believe increasingly powerful AI could potentially wipe out humanity.
“We really do earnestly believe AI could kill all humans!” Hubinger wrote. He estimated the risk of such an outcome within the next decade at more than 10%.
Altman says safety must come before further AI progress
Altman has also indicated that OpenAI is considering slowing development of its most advanced AI systems.
According to Bloomberg, Altman told OpenAI employees that the company was considering slowing work on its most cutting-edge AI and could potentially coordinate a pause with other AI labs. It was unclear whether other companies would agree.
Speaking to Fortune, Sam Altman said AI companies had a responsibility to put safety ahead of commercial incentives.
“We all have a tremendous amount of responsibility, and cannot let egos or incentives for profit or anything else get in the way,” he said.
Altman said a 10% risk of a catastrophic AI outcome was unacceptable and argued that more safety work was needed before pushing the technology further.
“I don’t think we’re currently at a place where we could say, you know, push much further on capabilities without making more progress on monitorability, alignment, the ability to understand what a model is doing, and the ability to make sure that a model will follow human values and the intent of its users,” Altman said.
Could AI companies legally coordinate a slowdown?
It remains unclear what a coordinated slowdown would involve, whether it would mean a temporary pause or limits on specific areas of AI development, and how compliance would be monitored.
There is also a legal obstacle in the US. AI executives have previously warned that coordination between competing companies could raise antitrust concerns.
“Some forms of coordination that would be impactful for pacing are legally challenging, and will require government support,” Amodei wrote in his Saturday blog post.
Altman and Amodei are both scheduled to appear at Salesforce’s Dreamforce conference in San Francisco next week. Altman is due to speak with Salesforce CEO Marc Benioff on Tuesday afternoon, while the timing of Amodei’s appearance remains unclear.
For now, Altman’s comments do not confirm that an AI pact is in place. But coming after Dario Amodei’s call for a slowdown, they signal that coordination among some of the industry’s biggest players is increasingly being discussed.
Leave a Reply