Trump bets on self-control by the tech industry in AI

After weeks of international debate on AI risks, US President Donald Trump has confirmed a clear course: no legal regulation – instead, Washington trusts the tech industry itself. On 30 September, leading corporations signed a voluntary self-control agreement at the White House.

Who sat at the table?

Among the participants of the meeting in Washington were Elon Musk (SpaceX), Mark Zuckerberg (Meta), Dario Amodei (Anthropic) and Jeff Bezos (Amazon). According to Trump, the group agreed on "a high level of self-regulation" – the text of the agreement is "morally binding", the President said.

What exactly was agreed?

Meta chief Zuckerberg described the arrangement as several layers of control mechanisms – including an internal risk assessment as well as external auditors and experts: "This is a first step that the entire industry could agree on." Trump mainly trusts in the economic interests of the companies: "Their companies are on the line if something goes wrong. They will not allow that."

Why the debate has escalated

The background is the repeated incidents at OpenAI (see block "OpenAI pauses training"), the warnings by Anthropic chief Amodei to slow down the development of more powerful models, and the appeal by 22 heads of state and government for international oversight. On 20 September, Trump had also announced an "AI Force" to monitor AI, and defends his line with a clear sentence: "Whoever wins in AI wins overall."

Assessment

Without legal binding, without independent oversight and without enforceable duties, the agreement remains at its core a gentleman's agreement. Observers consider the step a first – but widely insufficient – compromise between innovation pressure and safety concerns. How the EU and China react will be decided in the coming weeks.

Sources: tagesschau.de, 30.09.2026 · tagesschau.de, 22.09.2026