Ungoverned AI Is a Trillion-Pound Risk Surface. Here Is Where Governance Pays.

Adoption is moving faster than governance, and the risk has caught up. In the 2026 Allianz Risk Barometer, AI jumped from tenth place to second, the biggest riser in the survey. The question for risk owners is no longer whether AI is a material risk, but how large it is and where to spend against it.

Our analysis sizes the exposure across thirteen categories, from data breaches and algorithmic bias to agentic failures, deepfake fraud, shadow AI, and critical infrastructure. Each carries a per-event cost range and an annual aggregate. The total comes to between £1.6 trillion and £4.4 trillion globally by 2026, with the upper bound driven by the tail-risk profile of critical infrastructure.

The headline number is not the useful part. The useful part is leverage. Some risks respond strongly to governance investment, such as regulatory non-compliance, agentic failures, and critical infrastructure. Others are driven by external adversaries, such as deepfake fraud, or by employee behaviour, such as shadow AI, and respond far less. Spending evenly across the exposure surface wastes effort.

The return is real where the leverage is high. Moving from 30% to 70% governance maturity cuts aggregate expected annual loss by roughly 38% to 52%, and the estimated return is £8 to £22 of avoided loss per £1 invested. The first increments matter most, because they address the failure modes that are easiest to remediate.

Our whitepaper sets out the full model, the two sensitivity tables, and the leverage ratings, so risk owners can match investment to the categories where it works hardest.