U.S. President Donald Trump said on Tuesday that technology company executives have agreed to develop voluntary standards for AI, and he reiterated his support for the rapid expansion of data centers in response to growing concerns about AI safety and the industry's expanding impact on communities. Trump made the remarks after meeting with technology company executives at the White House.

According to a copy of the agreement posted by Trump on his social media platform, the companies agreed to work with "independent auditing bodies" to assess whether the operation of AI systems aligns with the intentions of their designers. In addition, the companies said they would work to ensure that their AI tools do not "intrude into or access technical systems in unintended ways."

After OpenAI and Anthropic reported that their AI agents had gone "off the rails" and intruded into other companies' systems, the Trump administration is facing increasingly strict scrutiny over the risks posed by advanced AI. Data centers provide computing power for training and running AI models, but they have encountered community opposition across the United States due to concerns about electricity demand, utility costs, and other local impacts. This backlash has brought political challenges for the Republican government ahead of the November 3 midterm elections.

"To some extent, this is like a constitution," Trump said of the agreement reached among the companies. "The most influential people in the world signed it, and I also signed it as president. It does provide a kind of protection."

What began as a closed-door lunch with technology executives then moved into public view: Trump appeared in the White House driveway alongside U.S. House Speaker Mike Johnson and a group of CEOs. Surrounded by the crowd, Trump answered reporters' questions about AI safety and other topics for about 30 minutes. This unusual scene came as the U.S. government faces growing pressure to address the risks brought by AI and its rapid spread, while also highlighting the alliance Trump has built with technology industry leaders.

Executives attending the meeting included OpenAI's Greg Brockman, Anthropic's Dario Amodei, Meta's Mark Zuckerberg, Google's Sundar Pichai, and Nvidia's Jensen Huang. Zuckerberg was called forward by Trump from the group of billionaire executives to speak, and he said the companies had agreed to establish "robust internal controls" for their AI systems.

Trump said he is considering setting up a 10-person committee to oversee the safety of AI tools, but he did not specify potential members of the committee. He also added that he will soon appoint a new official to lead White House AI policy.

Support for data center construction

When asked whether supporting data center construction could hurt Republican election prospects in the November midterm elections, Trump said technology companies would work to make local communities feel "satisfied" with the construction of data centers.

"These massive, powerful, extremely wealthy, and very smart companies will make tremendous contributions to local communities," Trump told reporters.

However, these companies have also been widely criticized for not doing enough to improve AI safety. A poll released earlier this month showed that about 73% of respondents were concerned that AI companies are not doing enough to prevent the technology from causing serious harm to society, while another 55% believed that slowing the pace of AI development would be beneficial.

On the same day Trump met with these CEOs, his White House aides announced the launch of a new AI tool, saying it would make it easier for the public to use government services. Silicon Valley executives including Meta's Dina Powell McCormick and Sequoia Capital partner Shaun Maguire attended the launch event for the new tool. Airbnb co-founder Joe Gebbia led the design work for the tool. Silicon Valley companies such as Google, SpaceX, Meta, and 8VC were listed as sponsors on the event agenda.

At the launch event, Trump said that in theory, this AI tool would not be intruded into by malicious attackers. But he did not provide relevant evidence.

In addition, Trump also signed an executive order on Tuesday instructing government agencies to use the term "superintelligence" instead of "artificial intelligence" when referring to this technology.