The AIHL Project presents research on POW deepfakes at UNIDIR

Nathan Heath and Dr Wm. Matthew Kennedy presented The AIHL Project’s research on deepfake threats to prisoners of war at UNIDIR’s Global Conference on AI, Security and Ethics in Geneva.

The conference took place on June 18–19, 2026, at the Palais des Nations and online, bringing together policymakers, researchers, industry, and civil society to discuss AI’s implications for international peace and security.

Synthetic media and humanitarian protection

Nathan and Matt’s joint session, “‘This black darkness of not knowing’: Can prisoner-of-war AI deepfakes be governed by IHL?”, formed part of the June 19 governance program. The 30-minute thematic deep dive examined synthetic media and the protections owed to prisoners of war under international humanitarian law.

The AIHL Project, which Nathan and Matt co-founded as a separate collaborative initiative, examines how those protections depend on information people can trust. Records help establish a prisoner’s identity and circumstances. Messages sustain contact with family. Images and recordings can become evidence of treatment in captivity.

Generative AI creates ways to falsify that information and to cast doubt on authentic material. The project studies these risks across the systems that create, handle, and distribute records, connecting technical safeguards with humanitarian responsibilities.

Connecting research and governance

For Syntony, this collaboration is part of a broader research agenda: examining specific ways AI can harm people, testing the safeguards available, and translating the findings into practical governance.

UNIDIR has published the conference agenda and recordings, including the second day’s governance sessions. Further information about the research is available through The AIHL Project.