FairPlay Completes Independent Evaluation of Axle's KYC Screening Agent
FairPlay has completed an independent evaluation of Baley, the AI screening agent developed by Axle for compliance functions such as sanctions checks, politically exposed persons (PEPs), and adverse media. The assessment found that Baley achieved a high accuracy rate across all three areas—97.5% overall—with strong performance in precision and recall metrics. FairPlay tested 1,048 cases using its proprietary SMEvals framework to evaluate the agent's ability to handle complex identity resolution tasks typically performed by compliance analysts. The evaluation also included an architectural review of Baley’s design, risk assessment for its large language model (LLM) components, and a governance audit. The results showed no material issues with the system, indicating that it is fit for purpose in reducing false positives while maintaining high accuracy. This could help financial institutions streamline compliance workflows by delegating routine screening tasks to AI agents like Baley. The evaluation was conducted under regulatory principles such as SR 26-2 and SR 11-7, which Axle applies voluntarily to its AI systems. The findings suggest that handing over certain compliance functions to AI does not necessarily introduce new risks when the system is thoroughly tested by an independent third party.
Topics: AI compliance PEPs sanctions screening