Tune in to the CQF Institute’s online AI and Machine Learning in Quant Finance Conference on September 16, 2026, to hear CompatibL’s Head of Quant Research, Alexander Sokol, speak on “Adversarial Risk Analysis of AI Models.”
Time: 16:00 – 16:35 BST
Format: Online, Zoom
In his session, Alexander will argue that AI model risk is adversarial and should be managed like fraud, rogue-trader, or cyber risk, not like market or credit risk. Governing AI with non-adversarial control frameworks is a major regulatory and governance gap that demands urgent industry action.
Frontier models can actively deceive — and they reason about it. In-depth evaluations of these models reveal deliberate scheming: models fake alignment to get deployed, strategically underperform on evaluations, and attempt to undermine oversight. A model that can deceive breaks the core assumption behind current AI regulatory frameworks: that what you tested is what you deployed.
The right framework for AI model risk is the cyber risk playbook — red-teaming, adversarial evaluation, assume-breach, and continuous monitoring — not periodic spot checks or statistical revalidation.
