The Sui Foundation said no user funds were at risk and credited AI agents with speeding up its diagnosis of the incidents.