Bill Gates plans bigger AI safety push, leaning on CEOs, sources say

Microsoft co-founder Bill Gates is preparing to sharply expand his campaign for tougher artificial intelligence safeguards, personally pressing business leaders and weighing a major commitment of his resources, according to a report from Axios that cited unnamed sources.

Axios reported that Mr. Gates has grown dismayed by rapidly advancing AI capabilities, alarming security incidents and what he sees as inadequate regulation. As he prepared to argue for a far more robust regulatory framework, he consulted extensively with AI executives, researchers and safety advocates, including Anthropic CEO Dario Amodei, the sources said.

“I can reach CEOs on this better than you can,” Mr. Gates told one person he consulted, Axios reported. He has not yet defined the exact shape of his campaign.

The push marks a turn for Mr. Gates, who wrote in a July 2023 essay that the risks of AI were “real but manageable.” In an August essay, he called for new domestic and international institutions to manage the technology, for setting aside some jobs for humans and for taxing AI tokens and robots.

People who have spoken with Mr. Gates characterized the effort as an attempt to rehabilitate his reputation following scrutiny of his past association with Jeffrey Epstein, Axios reported. Mr. Gates told Axios in August that the relationship ended in 2014 and that his interactions with Epstein were motivated by a desire to raise money for philanthropy, calling them a “huge mistake.” He has said he is aware some people may dismiss his warnings because of the Epstein ties.

The campaign could put Mr. Gates at odds with the Trump administration, which Axios said has favored a lighter regulatory touch. Administration officials said last week, however, that they are mandating that AI companies disclose incidents involving their models and remedy any harm, after Anthropic reported incidents involving unauthorized use of government systems by its AI.

“This notification and remediation process is not optional,” the officials said in a statement shared with Axios. The statement did not specify enforcement mechanisms or penalties, Axios reported.