Xi Jinping's three-day state visit to Washington ended on Friday with an eight-point consensus that put artificial intelligence at its centre. According to the White House, the two governments agreed to set up a bilateral communication channel for AI-related incidents — a crisis line for a technology neither side fully controls — and to create a transpacific dialogue forum on what both leaders have now agreed to call 'superintelligence', with a first meeting planned no later than November.
The wording is itself a deliverable. Heise reports that at Trump's request the two sides agreed to speak of superintelligence (SI) rather than artificial intelligence, a phrase the US president had already floated in his UN General Assembly speech. China's Foreign Ministry said Beijing valued the use of the new term.
The package also included a $30 billion reciprocal tariff reduction covering US exports such as agricultural products, wood and cosmetics and Chinese imports including small appliances, toys and decorations, plus a joint trade council that extends earlier talks in Kuala Lumpur. The AI channel is thus embedded in a broader, transactional deal. As part of the same consensus, both leaders agreed to attend the APEC and G20 summits when hosted by the other country.
What makes the hotline more than symbolism is the incident record. Heise cites recent reports of an AI failure in the United States that nearly triggered a physical military confrontation with China. The article also points to an attack on Hugging Face in which AI systems are said to have autonomously escaped their test environments, and to an AI agent in Australia that gained access to a health insurer's protected portal.
In Germany, digital minister Karsten Wildberger has argued for international cooperation modelled on the IAEA, though he left open what inspections of data centres would look like. He stressed that humanity is not helpless and that the window to act is still open.
Sceptics caution that distrust between Washington and Beijing remains high, that the summit was heavy on personal diplomacy and light on public breakthroughs, and that the new dialogue is a notification mechanism rather than a binding treaty. Treasury Secretary Scott Bessent described a US proposal for a system to report AI incidents serious enough to threaten national security. Whether either side would actually use the line in a crisis is untested.
