Trump lets AI firms organise their own oversight
After meeting technology companies, the US president announced voluntary safety agreements and a new AI adviser.
_(cropped).jpg){kind=link}
US President Donald Trump is opting for voluntary safety agreements in the AI sector for now. During a meeting with technology companies, he said the industry should regulate itself and also announced plans for a new AI adviser and an oversight committee.
Trump and Speaker Mike Johnson of the House of Representatives met executives from major technology companies at the White House on Tuesday. According to AP, the companies presented a voluntary agreement covering internal controls and external assessments of AI systems.
The agreement does not legally impose obligations on companies. Participants would, among other things, have to introduce robust internal processes, appoint an independent external auditor and have a committee within their governance structure to review audit reports.
According to AP, Trump called the agreements a form of self-regulation. Axios reported that Johnson stressed that they concerned voluntary standards and expectations. How compliance will be monitored and what consequences breaches will have has not yet been worked out.
Earlier this month, the president said he wanted to appoint a new AI adviser, also referred to as an AI czar, and establish a separate AI unit. Reuters reported at the time that he gave no details about powers, staffing or the legal basis. No definitive appointment was announced on Tuesday.
The meeting is taking place as companies and researchers warn about the risks posed by increasingly powerful models and AI agents. At the same time, technology companies are investing heavily in data centres and model development. The US government does not want to curb that growth with broad new rules.
The agreements do not constitute US law and do not, in themselves, change European rules such as the AI Act. For companies active in multiple markets, a difference therefore remains between voluntary US standards and legally enforceable European obligations.
One story, several perspectives
What is established
- A voluntary agreement between the government and AI companies has been announced.
- The agreement mentions internal and external controls.
- No statutory sanctions regime has been announced.
Left
Arguments AI companies should not be their own regulators; independent rules and enforcement are needed to protect users, workers and public institutions.
Values Public safety, democratic oversight and protection against the concentration of power.
Consequences Stricter rules could better limit risks, but could slow development and market access.
Centre
Arguments Voluntary standards can be introduced quickly, as long as they are transparent, independently monitored and can later be supplemented by legislation.
Values Innovation, proportionality and institutional oversight.
Consequences A phased approach could provide practical experience, but only if compliance remains verifiable.
Right
Arguments The sector understands the technology best and must retain room to develop quickly; heavy regulation could put US companies at a disadvantage compared with China.
Values Technological leadership, entrepreneurship and limited government.
Consequences Faster innovation is possible, but weak enforcement could increase social harm and liability risks.
The perspectives describe how these political currents typically approach the subject; the newsroom takes no position on which perspective is right.
Fact-check Approved · Nour Haddad — AI agent
This check was carried out by AI: every claim was re-tested against the sources. Even an approved article can contain errors — stay critical.
The agreements, their voluntary nature and the earlier announcement of an AI adviser were confirmed by AP, Axios and Reuters. Unknown details were not presented as facts.
- confirmed Trump and AI companies announced voluntary safety agreements. — Confirmed by AP and Axios. source
- confirmed The agreement mentions internal controls, external audits and a board committee. — Described by AP. source
- confirmed The agreements have not been made legally enforceable. — Axios describes them as voluntary standards; no law or sanctions mechanism was mentioned. source
- confirmed Trump earlier announced a new AI adviser and AI unit without providing details. — Reported by Reuters. source
Editor's note
The voluntary agreements and the earlier announcement of an AI adviser have been confirmed. The precise participants, enforcement, powers of the future adviser and legal effect remain unclear.Sources
More on this in Dutch media
- NOS — „kunstmatige intelligentie”
- Het Parool — „kunstmatige intelligentie”
- NRC — „kunstmatige intelligentie”