Computer Says Kill: How To Say No W/ Matt Mahmoudi and Marwa Fatafta

The final episode of the Computer Says Kill series spotlights a newly released joint statement signed by over 250 civil‑society groups, scholars, and tech workers. It reacts to high‑profile incidents such as the Iranian school strike and the use of large‑language‑model translations to generate kill lists in Gaza, illustrating how generative AI has already been woven into lethal targeting. By framing AI militarization as a pressing human‑rights crisis, the hosts connect the abstract debate to concrete harms that affect communities worldwide, underscoring why policymakers and businesses must act now.

The statement lays out four core arguments. First, it calls for an immediate halt to the provision of AI systems for any part of the military kill chain, arguing that speed, opacity, and entrenched bias make lawful use impossible. Second, it rejects the myth that a human‑in‑the‑loop or better data can render these tools compliant with international humanitarian law. Third, it places responsibility on the entire AI supply chain—model developers, data aggregators, cloud providers—to conduct UN‑guided human‑rights due diligence and refuse contracts that risk complicity in war crimes. Fourth, it documents documented abuses, from biometric surveillance in occupied Palestinian territories to AI‑driven targeting in Iran, to prove that the technology is already facilitating grave violations.

For a professional audience, the implications are clear: ignoring AI’s role in warfare exposes companies to legal liability, reputational damage, and moral failure. The statement urges firms to stop selling or exporting decision‑support AI for lethal or even non‑lethal military purposes unless transparent oversight and genuine human control are guaranteed. States are likewise asked to enforce international law and demand accountability from domestic tech firms. By aligning corporate policy with the UN Guiding Principles on Business and Human Rights, businesses can protect themselves while contributing to a broader effort to keep algorithms out of the battlefield.