Anthropic He suggested on Thursday (June 4) that developers of frontier AI should slow down or pause their efforts and give societal structures research time to keep up with the advancing technology.
the Anthropic Institute It will organize conversations with other AI companies as well as policymakers, researchers and civil society to address concerns about the technology and create ways to coordinate, Anthropic Institute leader Marina Favaro And co-founder Anthropy Jack Clark Written on Thursday Blog post.
“A meaningful slowdown or pause would require multiple well-resourced laboratories at or near the border, in multiple countries, to agree to stop under the same conditions,” they wrote. “It will also require each party to be able to verify that the others have actually stopped.”
The pause proposal was prompted by the increasing share of AI development being delegated to AI systems, and the possibility that one AI system could design and develop another AI system completely independently, according to the post.
The authors wrote that this development is not inevitable, but it may come sooner than expected.
“Artificial intelligence capable of constructing itself would be a major development in the history of technology, one that could bring enormous benefits to the world in science, health care and beyond,” Favaro and Clark wrote. “But full recursive self-improvement may also increase the risk of humans losing control of AI systems.”
Advertisement: Scroll to continue
The post came two days after President Donald Trump signed an executive order seeking access to new AI models up to 30 days before they are released so the government can choose “trusted partners who will have early access to covered coverage.” Boundary models To promote safe innovation and enhance cybersecurity of critical infrastructure.
It also came about a week after the Illinois governor’s announcement. JB Pritzker He said he would sign a bill requiring large border developers to create one Limits of artificial intelligence A framework that addresses risk and governance; Requires transparency reports before publishing boundary models; Mandates annual third-party audits; Requires frontier developers to report serious safety incidents; It provides whistleblower protection and internal reporting processes for covered employees.
