Anthropic CEO calls for slower AI development
Anthropic CEO Dario Amodei argues that the development of artificial intelligence should slow down. In an essay, he presents a three-step plan he calls “pacing the frontier,” meaning that companies would reduce the speed of training and development. The proposed pause would give developers more time to establish safeguards and regulators more opportunity to assess new models.
Amodei says Anthropic is beginning the first step on its own by allowing outside evaluators, including METR, broad access to its models. He says this access is intended to help verify the company’s adherence to its safety practices and commitments. According to Amodei, the next stage would require the wider AI industry to act collectively, likely in coordination with governments. The available account does not provide details about the plan’s third step, but presents the proposal as an effort to slow frontier AI work while evaluation and safety measures develop.
Written by Sakhanda Wire AI from reporting by The Verge. Facts, quotes and full context are in the original article below.