Zuckerberg: AI companies incentivized toward safety

Meta CEO Mark Zuckerberg on Tuesday argued that artificial intelligence (AI) companies are incentivized toward safety when it comes to the timing of crafting their models and that external pressure is not needed to keep to that timeline.

“Every lab has the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens,” the tech mogul wrote in a post on the social platform X.

Zuckerberg also asserted in his Tuesday post that “people won’t want to use agents that are misaligned with them and that don’t do what they ask,” putting pressure on AI creators to align models more with human values.

“There is a lot of debate about slowing progress on capabilities until alignment catches up,” he added. “My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models. Any lab that doesn’t focus on alignment will fall behind.”

Zuckerberg’s post follows a larger debate in the past week over AI development, with Anthropic CEO Dario Amodei saying over the weekend that development of the technology must be slowed to give space for the implementation of necessary risk prevention safeguards. Fellow tech moguls Sam Altman and Elon Musk backed Amodei up.

“We must slow the pace at which we improve the capabilities of AI models,” Amodei wrote in a blog post. “Progress will still seem fast, and we must make wise use of the time we gain.”

The Anthropic CEO recommended that AI companies quickly undertake measures aimed at “pacing” the development of models and warned of potentially grim consequences if the technology does not have brakes.