After a series of incidents in the information security field, industry representatives almost simultaneously began to express the idea of regulating the development of artificial intelligence. Anthropic CEO Dario Amodei has called on artificial intelligence startups to slow down, a move that has received preliminary approval from rivals OpenAI and SpaceX (xAI).
Image source: Human
At the very least, Amodei’s idea was unusually quick to gain support on social networks from Sam Altman and Elon Musk, which is not entirely typical for the current competitive AI industry. Altman explained that the startup decided not to IPO in 2026 due to the security situation of AI, but postponed it to a later time.
Dario Amodei, writing on the Anthropic blog, lays out a three-step plan to curb the pace of artificial intelligence without hurting the commercial interests of new startups or sacrificing U.S. leadership in the field. Some of the measures he proposes will be more difficult to implement, some will be easier, but industry will simply be forced to regulate itself.
First, Anthropic’s leaders proposed letting independent experts review internal security procedures and incident reports. The second phase invites leading companies from countries that consider themselves democracies to coordinate their work with each other and develop common safety standards. The third point of the plan invites countries with different political systems to come to an agreement on the development of artificial intelligence. Anthropic management publicizes such proposals in part to ease public tensions that have been stoked by revelations from experts who have left Anthropic and OpenAI, who accuse startup management of ignoring human life.
Amodei explained that ideas about slowing down the development of artificial intelligence have been circulating since 2023, but then they were not very relevant for implementation because at that time the capabilities of the artificial intelligence models were not enough to pose a serious threat to humanity from a practical perspective. “I still believe that artificial intelligence has the potential to greatly improve the quality of human life. My desire to achieve this goal has not diminished, but these benefits can only be realized if we develop the technology in the right direction, and given the use of time to make progress, special care must be taken to get it right. “ – Amodai admitted in an article on the Anthropic website.
Faced with all this, the company’s chief insists that the best pace of development needs to be chosen, because if the United States develops artificial intelligence too slowly, geopolitical rivals may take the lead. Sarah Heck, head of public relations at Anthropic, called on the U.S. government to join the industry regulatory process, ban the export of advanced chips to China, and pass a law requiring the selection of artificial intelligence models based on safety standards and mandatory preliminary testing.
Sam Altman expressed his support for Amodei’s views and explained that similar discussions have been held within OpenAI for several weeks. He supports the idea of competitors providing opportunities for third-party experts to work on advanced artificial intelligence models, and said he is ready to implement such measures on OpenAI. The company’s chief researcher Jacub Pachocki also recently called on market participants to voluntarily slow down the pace of artificial intelligence development until universally accepted safety standards emerge. Elon Musk, the head of xAI, a new startup owned by SpaceX, succinctly supported the head of Anthropic on the social network X page, admitting that he was right.
“I believe that even if the economic slowdown allows us to buy a year or two before models reach critical capability levels and use that time to get things in order, we can significantly reduce the risk of something going terribly wrong.” – noted the head of Anthropic.
If you find an error, select it with your mouse and press CTRL+ENTER.
