Acting Faster than the Threat: Protecting Critical Infrastructure in the Age of Frontier Models
Frontier AI does not create entirely new cyber threats; it destroys time. As advanced models compress vulnerability analysis and exploitation from weeks to minutes, critical organisations must align the speed of defence with the speed of attack, while keeping humans in command of high-impact decisions.
© Shutterstock
The principles of cybersecurity have not changed, but the grace period has disappeared. Frontier models can analyse code, identify vulnerabilities, demonstrate exploitability and generate attacks faster than human teams. Technical debt therefore becomes a business-continuity risk, automated attacks make organisations of every size viable targets, and strategic autonomy becomes an operational necessity.
Thales sets out a pathway from periodic, fragmented defence to continuous, integrated and governed cyber defence. Its approach combines a cross-functional framework for assessment, governance and AI FinOps with five pillars: internal applications; detection and response; a global cyber defence platform; continuous testing and risk assessment; and protection and resilience.
The white paper presents a roadmap to stabilise and decide within 45 days, build the foundations within 90 days, and move towards continuous defence over 12 months, with automation operating under human authority.