World 'may not have time' to prepare for AI safety risks, says leading researcher

Open in new window