Autonomous cyber defence for military systems using cognitive AI agents

The Challenge

Military operational platforms and networks face cyber attacks at machine speed, in environments where skilled cyber personnel are scarce and connectivity back to central security operations cannot be guaranteed. Dstl’s research agenda — including its Autonomous Resilient Cyber Defence innovation focus — seeks self-defending, self-recovering concepts: systems that can detect, decide and act against attacks autonomously, reducing the burden on human operators.

The Solution

For Dstl, NquiringMinds developed ACoSA — Autonomous Cognitive Security Agents: autonomous, resilient cyber-defence intelligent agents capable of cognitive decision-making for defending systems in defence and security contexts. The work drew on NQM’s trusted-data and AI capabilities to give agents the situational data, inference and policy machinery needed to reason about threats and respond without continuous human direction.

Outcomes

The work was delivered in full over twelve months (July 2022 – June 2023) with Dstl technical monitoring (Bethann Coleman, Duncan Hodges), and sits within NQM’s broader defence AI portfolio alongside MAXAD and BAE INTeACT IAMD, where the same autonomous, trust-scored data techniques are applied to naval combat systems and defence supply-chain capability.

TODO: confirm formal deliverables and outcomes.

In Detail

Autonomous cognitive agents that defend military systems against cyber attack.

Military platforms and networks face cyber attacks at machine speed, in environments where skilled cyber personnel are scarce and connectivity back to central security operations cannot be guaranteed. ACoSA — Autonomous Cognitive Security Agents — delivers resilient cyber-defence intelligent agents capable of cognitive decision-making: they detect, decide and act against attacks autonomously, reducing the burden on human operators.

Core benefits

• Autonomous decision-making — agents assess threats and select defensive actions without continuous operator input.
• Machine-speed response — attacks are met faster than human-in-the-loop processes allow.
• Resilience — agents are designed to continue defending and recovering systems while under attack.
• Trusted reasoning — situational data, inference and policy machinery let agents reason about threats within defence security constraints.