Before the first AI crisis, Washington and Beijing need a ...

A piece published on securityandtechnology.org argues that Washington and Beijing need a habit of talking about AI risk before the first AI crisis. The piece's stated premise is that AI risk does not take a break when relations between the two countries sour. It argues that either the world's two leading AI powers manage these risks together, or no one does. The headline frames the goal as a habit of talking, rather than as a specific agreement or mechanism.

The piece says some of the hardest design questions involve labs, not governments. On incident reporting, it states that no one has settled the method or speed at which a frontier developer should report an incident to its own government. By the piece's own account, then, neither the procedure nor the pace of such reporting has been agreed. The unanswered questions it identifies sit with the developers of frontier systems as much as with the governments that would receive an incident report.

The piece presents the two powers as the actors on whom the management of these risks depends, given its claim that if they do not manage them together, no one does. It places the unresolved design questions inside the labs, rather than inside the government bodies that would receive reports. And it frames the whole exercise as something that should precede a first crisis rather than follow one. The habit of talking, on the piece's account, is meant to be in place before the event it is intended to address.