Financial regulators have spent the past three years studying AI in financial markets. They are no longer studying. They are acting. Here is what is actually changing in 2026 and what it means for firms deploying or planning to deploy AI in risk, compliance, and investment functions.
The EU AI Act: Financial Services Implications
The EU AI Act classifies certain AI systems as high-risk, with correspondingly stringent requirements for transparency, explainability, human oversight, and auditability. Several categories directly applicable to financial services fall into this classification: AI systems used in credit scoring, insurance underwriting, and systems that materially affect employment decisions.
For asset managers and banks operating in the EU, the practical implications are: mandatory conformity assessments before deployment, technical documentation requirements, post-market monitoring obligations, and incident reporting for high-risk system failures.
The compliance window is not indefinite. Firms that have not begun conformity assessment processes for their AI systems should treat this as a priority, not a roadmap item.
SEC AI Guidance: The Conflicts of Interest Framework
The SEC's updated guidance on AI in investment advice focuses on conflicts of interest — specifically the risk that AI systems optimise for outcomes (fees, engagement, product sales) that are misaligned with client interests. The guidance requires disclosure of any AI systems used in the generation of investment recommendations and mandates that firms test their systems for bias toward conflicted outcomes.
For wealth managers and RIAs, this is a direct compliance requirement. For institutional asset managers, the implications are more subtle but present: AI systems that influence trade allocation, product selection, or research distribution need to be reviewed through this lens.
Basel Committee Discussion Paper: Credit Risk Models
The Basel Committee's discussion paper on AI in credit risk models doesn't create immediate compliance obligations, but it signals where the requirements are heading. The key themes: supervisory expectations for model interpretability will increase, statistical validation requirements will be extended to cover distributional shift and performance under stress, and internal model approval processes will need to account for the dynamic nature of ML models.
The window for proactive engagement with regulators on AI is open now. Supervisors in the US, UK, and EU are actively engaging with firms that approach them transparently. This is not a window that will remain open indefinitely.
What Firms Should Do
The practical agenda for financial institutions in 2026 is: inventory your AI deployments and classify them against the regulatory frameworks now in force; conduct bias and fairness assessments on any system touching credit, employment, or investment advice; establish audit trail requirements for all AI-generated outputs; and build the internal expertise to engage with regulators proactively rather than reactively.
We cover the full regulatory landscape — US, UK, and EU — in the June 10 intensive, with practical implementation frameworks developed from direct engagement with supervisory bodies. For risk and compliance professionals, this section of the course alone justifies the investment.
Take the next step — in a single day.
Everything discussed in this article — and the practical frameworks to implement it — is covered in depth in the June 10 one-day intensive with Irene Aldridge. In-person NYC from $1,295. Live webinar from $595.