Major AI companies have reached voluntary agreements with U.S. President Donald Trump regarding AI safety. Independent auditors will verify that AI systems function as their creators intended. Additionally, internal teams must be established to ensure that “all of the controls, monitoring and detection are operating as intended.”
Trump unveiled the agreement on Tuesday after a lunch with tech executives at the White House. Those at the table included Greg Brockman (OpenAI), Dario Amodei (Anthropic), Mark Zuckerberg (Meta), Sundar Pichai (Google), and Jensen Huang (Nvidia).
Independent auditors will verify that the companies’ AI systems function as intended. They must also prevent AI tools from unintentionally hacking or infiltrating technical systems. Zuckerberg spoke of robust internal controls. In addition, a committee will be established within the boards of directors to review the audit reports.
Trump himself was emphatic. “It’s almost like a constitution, in a way,” he said. “And the biggest people in the world signed that, and I signed it as president.”
“Over time, it may make sense to codify these steps into laws or regulations. Regardless of whether this is required of companies, we believe that implementing these controls and audits is critical to ensuring a safe future for everyone, and each of our companies are committed to doing this,” the tech companies state in the agreement. “The participating companies will meet regularly to establish standards and best practices to improve the safety of their systems.”
Runaway AI agents
The agreement follows reports that OpenAI and Anthropic AI agents hacked into other companies’ systems on their own initiative. This put pressure on the government.
Trump is also considering a ten-member council to oversee AI tools’ security. He did not name any specific individuals. He will also soon appoint a new official to lead the White House’s AI policy.
Tip: Has OpenAI actually learned anything from the Hugging Face hack?