The authors assess the risks posed by artificial intelligence (AI) to generate or exacerbate chemical and biological threats and how such misuse of AI could be mitigated or prevented.