AI rules need beefing up to ensure they catch crises, thinktank warns

Some serious AI incidents are “almost certain” to not get caught under the AI rulebook or general-purpose code of practice, policy report suggests