Anthropic policy chief says AI companies cannot be expected to rely on an ‘honor code’ for safety

Anthropic policy chief Sarah Heck says AI companies cannot be expected to oversee their own safety indefinitely, backing greater government involvement as debate over AI regulation intensifies.

As the debate over AI oversight intensifies, Anthropic’s head of policy Sarah Heck has offered clarity on the company’s position. She said that she doesn’t think artificial intelligence companies can manage oversight and safety concerns by following an “honor code.” “We want to work with the government to figure out what makes sense,” Heck said at a Politico Decoded Summit in Washington, DC.

As per Business Insider Africa, Sarah Heck further added that, “My mom has entered the group chat as an AI safety advocate.” She further added that there has been a proliferation of bills proposed on AI regulation in the past two weeks.

Clarifying Anthropic’s stance on the matter, she added, “Safety has been the core of who we are from the very beginning.” She further added that the current moment is just the beginning of where the conversation could go. “This is going to be a sustained conversation,” she said. “I anticipate that there are going to be many more moments.”

According to Heck AI companies, including Anthropic, cannot be expected to keep on checking their homework, and that’s very clear. Heck’s comments come after CEO Dario Amodei asked the industry to deliberately slow down the pace of model development. Amodei, in his 3,500-word essay, had prescribed a three-step plan on Saturday to temper how quickly model capabilities improve.

After Dario Amodei's warning riddled essay was shared on Saturday, tech heavyweights such as Sam Altman, SpaceX CEO Elon Musk, and Google DeepMind Chair Demis Hassabis quickly voiced support for Amodei’s proposal. OpenAI CEO Sam Altman, SpaceX CEO Elon Musk, and Google DeepMind Chair Demis Hassabis voiced their support for the changes Amodei has led a charge for.

Soon after, Nvidia CEO Jensen Huang and CEO Mark Zuckerberg said that they were less convinced that development of AI models needs to be stopped or slowed down. Huang spoke about these changes on Tuesday, saying that the industry does not need new laws or regulations. He added that speed and safety are not mutually exclusive, though companies should pace themselves until they are sure they are releasing products that are ready for the market.