The use of artificial intelligence (AI) in counterterrorism is not inherently wrong. This paper suggests some necessary conditions for legitimate use of AI as part of a predictive approach to counterterrorism on the part of liberal democratic states.