The big problem behind Trump’s AI accord

President Donald Trump is determined to leave AI companies to their own devices when it comes to building safety guardrails. But tech executives and former government officials say they’re skeptical the accord will lead to tangible changes because the stipulations aren't legally binding and likely aren’t much different from what the companies are already doing. The accord underscores a simmering tension between some major AI companies and the Administration over safety guardrails: Executives such as Anthropic CEO Dario Amodei have called for more substantive government regulation, but the Administration has not pursued such measures. Senate Majority Leader John Thune (R-SD) said that legislation will be needed “at some point,” though he said the accord was “good progress.” Instead, the White House has repeatedly rejected government-imposed guardrails. In that vein, the new AI accord is largely voluntary: President Trump described it as “morally binding.” Asked why it is not legally binding, the White House pointed to Trump’s press gaggle in which he said, “I’m seeing tremendous self-policing, and they understand that they have to self-police.” Alex Jacquez, Groundwork Collaborative’s advocacy and policy chief, said that the voluntary terms threaten to lull the government into thinking that the AI safety issue has been put to rest.