Four major giants of the artificial intelligence industry, who rarely agree on anything, set aside years of competition and disagreement within a nine-hour span on Saturday and all supported slowing the development of their models.
According to ITNA quoted by Axios, the companies' new position is that safety should be prioritized even at the cost of slower growth.
Background
The relentless growth of artificial intelligence is these days met with increasing public reaction, some governments, and even the companies that build the technology themselves. Data centers face public opposition, concerns about catastrophic AI scenarios have increased, and with only 51 days left until the U.S. midterm elections, lawmakers have concluded that tighter regulation of the industry could win them votes.
A turbulent day of events
In that environment, the most powerful CEOs in the industry took a coordinated action on Saturday:
At 10 a.m. Eastern Time, Dario Amodei, CEO of Anthropic, published a 3,800-word essay warning that model development should proceed at a controlled pace, otherwise scenarios such as a rebellious AI dominating the entire internet could occur within as little as six months.
An hour later, Elon Musk, CEO of SpaceX, surprised the tech community by posting on X and quoting Amodei's essay, writing that Dario is right. The reaction was notable because Musk, who chairs xAI (the new name of X.AI), had previously declined to join industry calls to slow their development pace.
At 12:30 p.m., Sam Altman, CEO of OpenAI, tweeted that he agreed with Amodei on the need to control the pace of progress at the frontier of the technology.
Altman also said in an interview recorded Friday and released Saturday that his company will not go public this year, because there is so much safety work remaining that they prefer to take that path as a private company.
At 7 p.m., Demis Hassabis, Nobel laureate and CEO and co-founder of Google DeepMind and Chief Scientist at Alphabet, posted that Amodei's essay points to the right path and, although its details need further discussion, he sees its overall direction as correct for confronting this critical moment.
Why this sudden convergence
It appears that publicly prioritizing safety over technological or financial growth is the industry’s only way to prevent legal suppression whose terms might be set by governments rather than companies themselves. U.S. lawmakers have asked the Speaker of the House to reconsider during the congressional recess and take action on the AI threat; at the more extreme end of the spectrum, Senator Bernie Sanders has called for a complete halt and criminalization of the most advanced AI systems.
However, observers say there are enormous financial interests at play in the industry that make a complete stop unlikely, and what will probably happen in practice is a choice between massive wealth accumulation and slightly less massive wealth accumulation, not a true halt.
The contentious point
Meanwhile, there is no sign of such convergence in China. Some influential Silicon Valley figures, including David Sacks, a former Trump administration official, have argued that slowing the technology’s progress in the U.S. would merely provide an opportunity for competitors with no safety concerns. In a post, Sacks challenged OpenAI and Anthropic to voluntarily slow their pace without waiting for government regulation, warning that otherwise the move would be seen merely as an attempt to evade the law or a political stunt.
Amodei himself acknowledged in his essay that a safety regime will only be effective if authoritarian governments also join it, and wrote that slowing the pace in democratic countries depends on the gap that U.S. companies have over state-affiliated Chinese rivals, and slowing too much could create a national security risk.
Conclusion
According to Axios, last Saturday may be recorded in history as the day AI’s course changed; the day the industry giants committed to restrain themselves. But whether these companies will truly agree on the exact meaning of “oversight” and “slowing down,” and whether they will honor their commitments, is a question whose answer must be seen in the coming months.