
LABScon26 Replay | Governing Just-in-Time Intelligence: AI, Dual-Use Capabilities, and the Limits of Guardrails
ICRC’s Mauro Vignati argues AI guardrails can’t fix a deeper problem: a 70-year-old distribution model now applied to dual-use cognitive capabilities.

In his LABScon 26 talk, Mauro Vignati, Advisor on Digital Technologies of Warfare at the International Committee of the Red Cross (ICRC), argues that guardrails cannot address the deeper problem with frontier AI. Advanced cognitive capabilities are now distributed through the same on-demand, as-a-service model that has shaped industry, and later cybercrime, for seventy years. That model was designed for industrial efficiency, not for technology this powerful.
Vignati traces the model back to Toyota’s post-war just-in-time production system, built on the principle that customers get only what they need and pay only for what they get. The IT needed to coordinate that system helped give rise to application service providers and eventually to software-as-a-service. Cybercriminals later adopted the same logic for crimeware and ransomware, and hacktivists for DDoS campaigns. In each case the people taking part needed no technical skill of their own.
Model-as-a-service differs from its predecessors because the capability is dual-use at the source, supplied by private technology firms largely independent of who the end users are or what they intend. Vignati explains why criminals who tried to build their own AI systems have largely turned to wrapping commercial ones instead, leaving the legitimate economy to carry the cost of training and infrastructure. He then draws on cybernetics’ law of requisite variety, the principle that any system regulating another must match its complexity, to argue that AI governance has no simple fix. In its place he proposes a layered framework for deciding what a model can do, for whom and under what conditions.
His warning is that as long as the industry keeps offering powerful general-purpose models, criminals and activists will continue to parasitise commercial services.
For policymakers, AI providers and threat researchers, this talk offers a historical lens on how offensive capability spreads and a starting point for thinking about how to govern it.
About the Author
Mauro Vignati is Advisor on Digital Technologies of Warfare at the International Committee of the Red Cross (ICRC). His work focuses on the militarisation of digital space and AI, the exploitation of the cognitive domain as a means of confrontation, and the impact of digital technologies in armed conflict on civilians, critical infrastructure and society. He has more than 20 years of experience in cybersecurity and previously spoke at LABScon 2022.
About LABScon
This presentation was featured live at LABScon 2026, an immersive 3-day conference bringing together the world’s top cybersecurity minds, hosted by SentinelOne’s research arm, SentinelLABS.
Keep up with all the latest on LABScon here.


