As many states, especially those with large and resourceful militaries, are exploring the potential of using artificial intelligence (AI) in targeting decisions, there is an urgent need to understand the risks associated with these systems, one being the risks of bias. However, while concerns about bias are often mentioned in the military AI policy debate, how it manifests as harm and what can be done to address it is rarely discussed in depth. This represents a critical gap in efforts to ensure the lawful use of military AI.

To help bridge this gap, Laura Bruun and Marta Bo from the Stockholm International Peace Research Institute (SIPRI) unpack the humanitarian and legal implications of bias in military AI. They show how bias in military AI is likely to manifest in more complex and subtle ways than portrayed in policy debates, and if unaddressed, it may affect compliance with IHL principles of distinction, proportionality, and, especially, precautions in attack.