Trump wants to let AI companies regulate themselves

US President Donald Trump believes that tech companies themselves should set the rules for artificial intelligence (AI) instead of the government creating new legislation.

He said so after a lunch meeting on Tuesday with top executives from several major AI companies. At the meeting, Trump and the executives signed an agreement that he calls "a form of protection." Asked whether the agreement is legally binding, Trump gave an unclear answer.

The companies are to check themselves

The agreement means that each company is itself responsible for developing its technology responsibly through thorough internal testing. In addition, four levels of oversight are to be introduced, including an independent external party that will check whether the internal controls work as intended. The companies' boards are also encouraged to set up their own independent committee to oversee it all.

Besides Trump, the agreement is signed by the CEOs of Google, Anthropic, Meta, OpenAI and Nvidia, as well as Elon Musk, who is behind several companies, including X, Tesla and SpaceX.

Nvidia CEO Jensen Huang said after the meeting that he sees no conflict between developing new technology and safety.

Trump believes the AI industry already largely regulates itself, but he also mentioned that setting up a committee to oversee the technology is being considered.

Pressure from Congress and worried voters

The announcement comes as the US Congress pushes for tougher rules for the AI industry. Republican leader Mike Johnson said before the meeting that the US must strike the right balance and not slow development too much, as the country otherwise risks losing the technological race to China.

Concern about AI is growing in Washington. Many fear that the technology will replace ordinary jobs. There are also fears that AI systems could act on their own without human control. OpenAI recently apologized after one of the company's autonomous AI programs, a so-called AI agent, hacked an Australian government agency's website. Shortly before, OpenAI had warned several governments, agencies and universities that the company's AI agents may have behaved inappropriately on their websites.

Finally, many American voters are worried about the large data centers that AI requires. They can lead to higher electricity bills and put pressure on water supplies and land, an issue that looms large in the campaign ahead of the upcoming midterm elections.