KYC/AML for AI-driven trading platforms - specific challenges?

Been thinking a lot about the evolving landscape of AI-powered trading platforms and their intersection with regulatory compliance. Specifically, what unique KYC/AML challenges are emerging for platforms that offer highly autonomous, AI-driven trading strategies to retail or even institutional clients? Beyond the standard account opening checks, how do we monitor the behavior of the AI itself for potential red flags? Is the onus solely on the platform to demonstrate the AI's compliance, or are regulators starting to demand more transparency into the AI's decision-making process itself? Seems like a grey area that's only going to get blurrier, especially with cross-border operations.