American and Chinese security experts are calling for new safeguards to
prevent artificial intelligence from triggering a military or nuclear crisis.

Their proposals include banning AI systems from independently deciding to use
nuclear weapons or attack nuclear command networks.

They also say humans should retain control over major military cyberattacks,
particularly those targeting critical infrastructure such as energy,
healthcare and finance.

The experts are also proposing a dedicated U.S.-China hotline for incidents
involving autonomous AI. They warn that malfunctioning or compromised systems
could act without authorization, leaving either country unsure whether an
attack was intentional.

The recommendations come ahead of planned government-level talks on AI.

Neither government has endorsed the proposals, although both countries
previously agreed that humans should retain control over decisions involving
nuclear weapons.