The AI rulebook just split in two

Intergovernmental technology rules now come in two flavors: ones drafted by the powers whose firms build the technology, and ones drafted by the powers the technology is built against. Both now have a permanent home, and they do not share a room.

The second kind opened in Shanghai on Thursday, sealed by 29 governments: ten African states, twelve Asian states, plus Belarus, Serbia, Cuba, Brazil, and Venezuela, with the UN Secretary-General in the room and the European Union and the United States absent. The DNA is Global South plus Russia and China: rules written by governments that argue extraterritorial tech rules are a form of pressure.

Russian Deputy Prime Minister Dmitry Grigorenko's launch line, that Russia "consistently advocate[s] for the establishment of transparent rules governing extraterritorial technologies," is the load-bearing sentence. Read forward it is a launch statement; read against the EU AI Act and the U.S. export-control regime, it is a counter-standards claim about who has standing to govern the same technology.

The mechanism is portable. Any convening power able to assemble 29 governments, seat a UN Secretary-General, and table a counter-rule can mint a parallel institution for the next frontier technology. The AI rulebook's shape over the next year will be set by two numbers: how many of the 29 ratify a charter with real standards, and whether Brussels and Washington treat the body as a foil or ignore it. The next frontier technology's rules will be written in two rooms, not one.