Technology and Science · updated 31m ago · 3 min read
Trump Says AI Leaders Signed A “Constitution” For Self-Policing, Calls It Morally Binding
Trump and major AI leaders signed a voluntary accord to self-police frontier AI. Trump described the pact as morally binding and like a constitution.
Whether the pact will be enforceable.
10 of 12 outlets skipped it: trump signalled a DOJ and FBI role in AI issues.
25outlets compared
Same story, two versions
tap a side to read it in full
Semafor
“The still-unreleased pact raises questions about enforcement — Johnson referred to ‘consequences’”Read the original ↗
Semafor spotlights enforcement doubts, while ABC treats self-policing as the accord's core promise.
A “constitution” for AI
Donald Trump said AI leaders signed a “constitution” to police themselves, and he called the self-policing agreement “morally binding.” House Speaker Mike Johnson said the accord is “voluntary” for the AI companies and described it as “a statement of principles” with “commitments that are voluntary on behalf of the industry.” Trump said he might form a “committee” and floated putting “maybe 10 people on that committee” to watch over the “whole enterprise.” Trump also said the industry will be “self-policing,” and he pointed to his head when asked whether the voluntary document requires more than “love.”
“Trump said that the self-policing agreement is ‘morally binding.’”
The meeting at the White House included Meta CEO Mark Zuckerberg, Elon Musk, Anthropic CEO Dario Amodei, Google CEO Sundar Pichai, Jensen Huang from Nvidia, Palantir’s Alex Karp and Shyam Sankar, Amazon’s Jeff Bezos, and Greg Brockman from OpenAI.
Four layers of controls
The White House accord titled “The White House Accord on Superintelligence: A Joint Commitment on Frontier SI Responsibilities” set out commitments for companies to implement “solid internal processes and controls” to keep systems operating as intended. The agreement described four levels of controls, including internal mechanisms to monitor model capabilities and behavior during training and deployment, internal teams with authority to verify those controls, and “evaluations conducted by independent external auditors or reviewers.” The accord also called for an independent board committee to supervise the controls and receive reports from internal and external teams, and it said the mechanisms should pay special attention to risks related to cybersecurity, biosecurity, and chemical threats. Trump described the document as a form of protection and said the leaders of the AI sector signed a mandatory “ethical charter” to regulate the sector, while other reporting said the accord does not have legal force.
Trump said he is considering forming a 10-member committee to oversee the AI sector, and he said the group would “officially rename it as "superintelligence."”
Renaming and next steps
Trump ordered federal agencies to stop using the expressions “artificial intelligence” and “AI” and to replace them with “superintelligence” and the acronym “SI,” formalized by an executive order. The order set a 60-day deadline for the presidential Science and Technology adviser to present a proposal aimed at establishing a federal definition of “superintelligence.” Trump said he would officially rename the technology because “it’s not artificial,” and he said the agreement was “almost like a constitution” after the White House luncheon. Zuckerberg told reporters the agreement was “a start and an accord that the whole industry could come to,” and he said it involved “building multiple layers of auditing and controls.”
The accord left open the possibility of turning the commitments into laws or regulations, and it said it recognizes the steps outlined are not codified into formal laws or regulations. Trump said Congress would “continue to guide the industry's development in a safe manner,” but he stopped short of announcing any action to enforce the principles. Trump also said he opposed guardrails and regulations on the technology or anything that would “stifle” its growth, while the agreement emphasized each company’s individual responsibility to ensure its technology “builds trust with customers and the public.”