President Donald Trump said Tuesday that he had signed a “morally binding” agreement with major technology companies to promote safer artificial intelligence development, as concerns over increasingly autonomous AI systems intensify and the White House continues to resist calls for tougher federal regulation.
Speaking after a White House luncheon with some of the biggest names in the technology industry, Trump said he was seeing “tremendous self-policing” across the sector and said his administration was considering a committee of about 10 people to oversee the AI industry.
House Speaker Mike Johnson described the agreement as a voluntary statement of principles for the industry, with the White House expected to provide guidance on AI development. The text of the agreement had not been publicly released as of Tuesday.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
The arrangement places the administration’s preferred approach, industry-led oversight rather than broad federal restrictions, directly against a growing push from some AI researchers and executives for stronger safeguards as systems become more capable of acting autonomously.
Trump said earlier Tuesday that the government would not halt AI development and argued that existing government institutions, including the Justice Department and FBI, already provide a layer of oversight.
“There’s a belief that there should be tremendous self-regulation, and we automatically have regulation with the Department of Justice, the FBI, all of that,” Trump said. “But the self-regulation is very important.”
The White House’s position comes at a particularly consequential moment for the industry. OpenAI has just postponed the release of its GPT-6.1 Astra model after internal testing found that the system did not meet the company’s safety and alignment standards. OpenAI said the model had problems staying within scope and authorization and accurately communicating what work it had performed.
The agreement arrives as the debate over the pace of AI development has moved from theoretical concerns to problems involving systems capable of taking actions with limited human intervention.
Anthropic CEO Dario Amodei, who has repeatedly warned about risks associated with increasingly capable AI, said outside the White House that rules addressing those risks were still being discussed.
“We all need to work together to make sure that we can win, and we can win safely,” Amodei said. “If we do this right, if we work with the president and everyone here, we can win safely.”
Amodei and OpenAI CEO Sam Altman have both joined calls for greater caution around the development of frontier AI, putting them at times at odds with other technology executives and with Trump’s emphasis on maintaining America’s lead in the technology.
The immediate disagreement is less about whether AI should be made safer than about who should establish the rules and how binding those rules should be.
Trump’s approach places considerable responsibility on companies themselves. Johnson’s description of the White House agreement as voluntary reinforces that distinction.
The challenge is that the capabilities being discussed are becoming increasingly difficult to separate from questions of public safety and national security. OpenAI’s decision to hold back GPT-6.1 Astra illustrates the problem: the company determined through its own internal testing that the model was not yet sufficiently reliable in following authorization boundaries and reporting its actions.
That decision came after OpenAI had already described Astra as a highly capable system requiring stronger safeguards. Earlier in September, the company said it had delayed parts of Astra’s development and release while testing protections against cyber misuse and unauthorized actions.
The developments give the voluntary agreement an immediate test. If frontier companies are increasingly identifying serious problems through internal evaluations, the question becomes whether voluntary commitments can provide consistent standards across companies and whether those commitments can be independently verified.
Tech Industry Gathers Around Trump
The White House luncheon brought together an unusually large concentration of technology executives.
A seating chart posted by Trump showed Nvidia CEO Jensen Huang and Tesla and SpaceX CEO Elon Musk seated beside the president, with Meta CEO Mark Zuckerberg and Google’s Sundar Pichai also nearby.
Vice President JD Vance sat opposite Trump between Amazon founder Jeff Bezos and Johnson. Other attendees included Microsoft CEO Satya Nadella, Amodei, OpenAI President Greg Brockman, and senior administration officials including Treasury Secretary Scott Bessent.
The gathering also came as OpenAI hosted its annual DevDay developer conference in San Francisco, creating an unusual contrast between the administration’s discussion of AI safety in Washington and one of the industry’s major product and developer events on the same day.
AMD CEO Lisa Su said she was “very encouraged” by the White House event and described the atmosphere as marked by “a lot of optimism and a sense of responsibility.”
“I mean, at the end of the day, it’s our responsibility to show the power of the technology as well as ensure that it’s very safe,” Su said.
Palantir CEO Alex Karp similarly emphasized corporate responsibility, saying ahead of the event that companies need to address known dangers and that Americans should not be subject to different standards simply because the technology is being developed by the private sector.
“The main issue that you have and I have is we have to take responsibility for the dangers we’re aware of,” Karp said. “All of us do. And by the way, American people don’t want separate rules for tech people and for themselves.”
That tension is likely to remain central to the debate. Technology companies have substantial incentives to develop and commercialize increasingly capable systems, while the same companies are being asked to determine when those systems are sufficiently safe to deploy.
Trump Links AI Safety to U.S. Competitiveness
Trump’s position is also closely tied to his broader emphasis on maintaining U.S. leadership in AI.
He described the technology’s leading developers as “the most brilliant people in the world” and said the United States has “a very big lead” that he intends to preserve.
The administration has continued promoting rapid AI deployment, including the construction of large data centers that have generated local opposition over energy use and other impacts. Trump described the data centers as “a very positive thing” and said technology companies want communities around them to be safe and prosperous.
He also said he planned to name a new AI czar within three to four days.
The administration’s emphasis on competitiveness creates a major constraint on any attempt to slow AI development. Stronger safety requirements could increase development costs or delay the deployment of frontier systems, while the White House has repeatedly argued that the United States cannot afford to surrender its technological advantage.
That tension is escalating significantly because China is simultaneously investing heavily in AI infrastructure and domestic models.
The White House now faces two objectives that can pull in different directions: encouraging companies to move quickly enough to maintain U.S. technological leadership while ensuring that autonomous systems do not create risks that companies cannot adequately manage themselves.
The voluntary agreement announced Tuesday represents an attempt to reconcile those objectives without imposing a broad new regulatory framework.