# Built for Lethality, Not for Law: Rebalancing Military Artificial Intelligence Toward Civilian Protection and Factual Integrity under International Humanitarian Law **By:** Chris Rogers **Published:** 2026-09-21T13:42:25.542Z **Source:** [LinkedIn](https://www.linkedin.com/posts/chris-rogers-857823299_built-for-lethality-not-for-law-rebalancing-share-7507796379101102081-BQ4d) --- # Chris Rogers’ Post [Chris Rogers](https://www.linkedin.com/in/chris-rogers-857823299?trk=public_post_feed-actor-name) Pleased to share my new article, "Built for Lethality, Not for Law: Rebalancing Military Artificial Intelligence Toward Civilian Protection and Factual Integrity under International Humanitarian Law," forthcoming in the Asser Institute Yearbook of International Humanitarian Law (YIHL), Vol. 28. Militaries are rapidly integrating AI across operations — but investments overwhelmingly prioritize lethality and speed. This article argues that gap can and needs to close. Drawing on my time at the DoD's Civilian Protection Center of Excellence, I examine four areas where AI could be invested in and designed to affirmatively advance civilian protection and IHL. The urgency has only grown with AI-assisted military operations in Iran and the dispute between DoD and Anthropic over military AI. The full paper is available on SSRN: [https://lnkd.in/gqDR3Kqx](https://www.linkedin.com/redir/redirect?url=https%3A%2F%2Flnkd%2Ein%2FgqDR3Kqx&urlhash=FNrf&trk=public_post-text) [ Built for Lethality, Not for Law: Rebalancing Military Artificial Intelligence Toward Civilian Protection and Factual Integrity under International Humanitarian Law papers.ssrn.com ](https://www.linkedin.com/redir/redirect?url=https%3A%2F%2Fpapers%2Essrn%2Ecom%2Fsol3%2Fpapers%2Ecfm%3Fabstract_id%3D7477958&urlhash=DtqD&trk=public_post_feed-article-content) This is an important angle. A lot of the conversation around military AI seems to focus on speed, targeting and capability, but there’s another side to it can the same technology actually help people make more careful decisions? And I think that brings us back to how these systems are deployed, not just how good the model is. If AI is going to help assess civilian risk, the people using it need to know what information it’s working from, where that data is going, and why it produced a particular recommendation. In some environments, that probably also means the AI needs to work securely where the sensitive data already sits, rather than constantly sending it somewhere else for processing. The model can be incredibly capable, but if the person making the decision can’t question or trace its output, we haven’t really solved the harder problem. Really interested to see more discussion around this side of military AI. Congratulations. This is the kind of debate we should be having rather than alarmist apocalypse stories