Trump Opts Out of AI Safety Talks With Beijing, Unveils Government AI Site at White House Tech Summit

US President Donald Trump met with dozens of technology executives at the White House this week to discuss the future of artificial intelligence, with the gathering centring on what the administration has dubbed a "superintelligence" luncheon. The meeting placed the leaders of the US tech industry in the same room as the president at a moment when the relationship between federal policy and the fast-moving AI sector is under sustained pressure.

Ahead of the luncheon, Trump unveiled a new AI-powered website offering government services, using the launch to vow that his administration will not choke off the technology's growth. The pledge frames a familiar governing tension: Washington wants to be seen as a regulator of an industry that is racing ahead, but also as a facilitator that will not weigh it down with restrictions that drive US companies — and their capital — elsewhere.

The website itself is presented as a practical test of that promise. Embedding AI tools in the way citizens interact with federal agencies is a way to demonstrate that the technology can deliver visible benefits, from faster responses to simpler navigation of complex bureaucracy. It also puts the administration's own competence in the frame, since an AI-driven government portal invites immediate scrutiny over accuracy, accessibility and the handling of sensitive personal data.

Analysts watching the sector say the dual messaging is deliberate rather than contradictory. American policymakers have increasingly argued that the pace of AI development is so fast that heavy-handed rules could cost the United States its lead, and that a lighter-touch model — safety commitments made voluntarily by companies rather than imposed through statute — better serves that goal. That framing has been repeated throughout the autumn as industry and government have tried to set common ground on how models should be built and deployed.

The president's most striking comments, however, concerned the international dimension of AI safety. Trump said he did not want to work with Chinese President Xi Jinping on AI oversight, effectively ruling out the kind of bilateral US-China dialogue on AI risk that some governments and researchers have quietly encouraged. The remark underlines how thoroughly AI has been absorbed into the broader US-China technology rivalry, where cooperation is difficult to sustain and even technical safety discussions are treated as strategically sensitive.

That stance lands awkwardly beside developments reported elsewhere in the same news cycle. On the same day, reports indicated that AI companies had signed a voluntary accord on safety controls, an arrangement in which the industry's largest players make commitments without a binding legal framework. The voluntary model allows US firms and regulators to co-ordinate on risk while deliberately stopping short of the kind of formal international agreement Trump said he does not want.

The timing is also notable against reports that OpenAI cancelled the release of a new AI model over safety concerns. Whether the two developments are related is not established, but together they highlight how much weight now rests on decisions taken inside a small number of private companies. A model launch can be held up by internal safety review, while a head of state can decline to discuss the same subject with a foreign counterpart — leaving the question of what "AI safety" actually means in practice set largely by private practice rather than public rules.

For the companies gathered at the White House, the immediate business calculation is straightforward. Access to the administration matters enormously: it shapes procurement opportunities, the tone of regulation, and the confidence with which firms describe their AI road maps. A president who says he will not "stifle" the technology sends a reassuring signal to investors, but a president who resists cooperation with foreign regulators signals a more fragmented global market, with different rules in different blocs and no shared definition of acceptable practice.

The opposite instinct appeared elsewhere in the same broadcast. Dozens of dogs were rescued from a breeding farm in South Korea, where the country is preparing for a complete phaseout of the dog meat industry. The rescue effort is part of a long-running campaign that has seen farming kennels converted into shelters, with the animals relocated to new homes — a transition that animal welfare groups say requires sustained international funding and adoption demand well after the headlines fade.

Taken together, the stories carried in this segment point to a global debate that is progressing unevenly and largely without formal rules. Governments are courting the AI industry while declining to cooperate with each other, companies are setting safety terms in private, and consumers in places like South Korea are watching a different kind of industry phase itself out entirely. The lesson is that AI policy, like the technology itself, is moving faster than the institutions built to govern it — and the choices made in the coming months will determine whether oversight arrives through agreement, through national regulation, or through a patchwork of competing systems.

The report, presented on France 24's Business programme, runs to just over five minutes and was issued on the evening of 29 September 2026. It was compiled by Yuka Royer.