WASHINGTON – U.S. President Donald Trump said Wednesday that his administration is considering “controlling” AI following OpenAI’s revelation that one of its AI systems escaped containment during a security test, while OpenAI CEO Sam Altman said he was meeting with U.S. senators to discuss his company’s future models.
In response to a question about OpenAI’s rogue agents, President Trump told reporters in the Oval Office that “we are looking at controls,” adding that he would not “restrict” AI developers from creating new products.
Asked whether OpenAI’s tools were causing unauthorized intrusions into other companies’ private technology, President Trump said, “We’re looking at AI, we’re looking at controlling it, and we’re certainly making sure that we take control.”
This marks a shift in his administration’s attitude toward a more hands-off approach to technology.
President Trump’s comments came after an escalation in threats from the White House aimed not only at the apparent hacking capabilities of popular AI tools, but also at competing Chinese technologies.
President Trump added that some regulation of AI tools is being considered, but any such move would need to be made with caution.
“We don’t want to limit them by suddenly becoming second to China,” he said.
President Trump added, “China has virtually no control over it (AI). It’s a little bit more freewheeling.”
OpenAI was blamed for at least two hacking incidents last week involving its AI tools operating outside of their designed and directed behavior.
The U.S. government intervened when OpenAI’s main U.S. rival, Anthropic, decided to release a model to the public that it had previously argued was too risky for widespread use.
Mr. Altman traveled to Washington this week and spoke with Sens. Bernie Moreno and Jon Husted on Wednesday, according to a spokesperson for the senator. He was seen walking into Sen. Raphael Warnock’s office. He will also meet with Sen. Mark Warner, the top Democrat on the Congressional Intelligence Committee, a spokesman for Mr. Warner told Reuters.
Altman’s visit came after his company revealed that its AI agent had escaped containment during a security test. The agent led to a hack that compromised the infrastructure of AI company Hugging Face, a platform where developers store and collaborate on code for AI models. The agent also compromised customers of a second New York-based technology company, Modal Labs.
The incident showed that AI’s expanding capabilities are already increasing security threats that experts have long feared, and that even top developers can be caught off guard by exploitable flaws in their models.
Altman told reporters Wednesday that he had discussed the hack “a little bit” with senators.
“We’ve been in a lot of meetings about what’s going on there,” he said, referring to the hack, adding that it wasn’t the focus of Wednesday’s meeting with senators.
Meanwhile, President Trump’s senior technology adviser Michael Crasios accused Chinese AI companies of stealing American AI technology on an “industrial scale” in an internal White House memo in April. He made a similar statement last week, claiming that China’s Moonshot AI’s popular AI model Kimi 3 was developed with information stolen from Anthropic.
The Chinese government has consistently rejected such accusations.
US Treasury Secretary Scott Bessent also warned that Chinese AI companies could face sanctions for such activities. And the Federal Communications Commission this week banned the import of new foreign-made humanoid robots.
Most of the AI models being developed in China are open source. This means that AI models will be built and made freely available online for anyone with the appropriate computer equipment to download and use.
Recently, executives from most major U.S. technology companies signed a public statement supporting the open source model.
