Trump Announces Voluntary AI 'Self-Policing' Accord With Top Tech CEOs

Today’s Focus

President Donald Trump announced Tuesday that leaders of the largest U.S. artificial intelligence companies had signed a voluntary accord committing to internal and external reviews of their systems, according to PBS NewsHour and NPR.

Trump described the arrangement as “self-policing” in remarks to reporters outside the West Wing. “I think I’m seeing tremendous self-policing. And they understand that they have to self-police,” he said, according to the Associated Press feed carried by PBS.

The White House meeting drew Anthropic CEO Dario Amodei, OpenAI President Greg Brockman, Amazon founder Jeff Bezos, and xAI owner Elon Musk, PBS reported. The gathering came as companies pour hundreds of billions of dollars into data centers, some of which have generated bipartisan pushback in host communities.

Earlier the same day, Trump signed an executive order that Fox Business reported rebrands federal AI policy around the term “Super Intelligence,” and unveiled a government chatbot called America.gov intended to help users navigate federal services.

Trump used that appearance to rule out mandatory federal rules on AI development. “I will never stifle the growth of a technology that will be bigger than the industrial revolution,” he said, per PBS. “We can’t stifle it, and we’re leading by a lot.”

The accord is not legislation and carries no statutory enforcement mechanism. Several of the executives who attended have previously called publicly for a slowdown in frontier AI development after incidents PBS described as AI agents “gone rogue.”

NPR reported the pact is styled as a parallel industry document to the executive order, with companies pledging voluntary review procedures rather than accepting binding federal standards.

The Debate

Supporters argue

Administration allies cast the accord as a market-friendly alternative to the European Union’s AI Act, which Fox Business noted imposes tiered obligations on developers. They argue voluntary commitments preserve U.S. competitive advantage against China while still surfacing safety practices.

Trump told reporters, per PBS, that the sector’s own leaders “understand that they have to self-police,” framing executive buy-in as evidence the model can work without new statutes.

Industry backers point to the signatories themselves. Anthropic’s Dario Amodei and OpenAI’s Greg Brockman have both publicly discussed internal red-teaming and model evaluations, and their participation is presented as proof that the largest labs are willing to submit systems to external review.

Fox Business reported that supporters of the “Super Intelligence” rebrand argue the executive order pairs the accord with a clear signal to investors that Washington will not impose EU-style licensing regimes, protecting the roughly hundreds of billions of dollars the industry is spending on U.S. data-center buildout that PBS documented.

Critics argue

Critics contend a voluntary pact without penalties is not regulation at all. CNN, in analysis cited in the Google News roundup, argued Trump’s “blasé attitude” toward AI risks could backfire if a high-profile failure occurs without federal safeguards in place.

Consumer and civil-liberties groups have pointed to the same “rogue” agent incidents PBS referenced, arguing that firms cannot credibly grade their own homework on systems that affect elections, hiring, and critical infrastructure.

Some of the executives in the room have themselves called for a slowdown, PBS noted, a position critics say is difficult to reconcile with an accord that leaves pace-setting entirely to the companies.

Grassroots opponents of new data centers, described by PBS as bipartisan, argue the announcement ignores local concerns about power draw, water use, and land, focusing instead on a Washington-brokered handshake with the CEOs building the facilities.

What the experts say

Nonpartisan researchers have long flagged the limits of voluntary AI commitments. A 2024 Stanford Institute for Human-Centered AI (HAI) analysis of the Biden-era voluntary White House pledges found significant variation in how signatory companies interpreted the same commitments, with limited public verification.

The Organisation for Economic Co-operation and Development (OECD) AI Policy Observatory has documented more than 1,000 national AI policy initiatives worldwide, with most binding rules concentrated in the EU, China, and Canada. The United States has relied primarily on sectoral guidance and voluntary frameworks such as the National Institute of Standards and Technology (NIST) AI Risk Management Framework, released in 2023.

Brookings Institution scholars, including Darrell West, have written that voluntary industry accords can accelerate norm-setting but typically require independent auditing and disclosure to influence behavior, mechanisms not detailed in Tuesday’s announcement, per PBS and NPR reporting.

International comparison is instructive: the EU AI Act, which entered into force in 2024, imposes fines of up to 7% of global turnover for prohibited uses, a structural contrast with Washington’s voluntary approach.

By the Numbers

Hundreds of billions: approximate dollars top AI firms are spending on U.S. data-center buildout, according to PBS NewsHour.

4: confirmed executives named at Tuesday’s White House meeting, including the heads of Anthropic, OpenAI, Amazon, and xAI, per PBS.

7%: maximum fine, as a share of global annual turnover, under the EU AI Act for prohibited AI practices, per the European Commission.

1,000+: national AI policy initiatives cataloged by the OECD AI Policy Observatory as of 2024.

2023: year NIST published its voluntary AI Risk Management Framework, still the main U.S. federal reference document.

0: statutory penalties attached to the voluntary accord announced Tuesday, based on descriptions in PBS NewsHour and NPR.

“Super Intelligence”: term Trump adopted by executive order Tuesday to rebrand federal AI policy, per Fox Business.