AI & RegTech for Financial Services and Insurance Summit

Banks, insurers, and FinTechs are getting mixed signals from Washington.  In January 2025, the Administration ordered agencies to unwind Biden-era AI risk rules, and deliver a national AI Action Plan.  The July 2025 plan emphasizes fewer federal constraints, faster data-center permitting, standards-led exports, and procurement guardrails for “frontier” LLMs.  Meanwhile, the Senate stripped a proposed 10-year freeze on state AI laws, leaving room for states like New York and Colorado to press ahead as the EU AI Act’s early obligations begin taking effect, inviting mounting obligations from states and abroad that financial institutions cannot ignore.  In this opening session, topics of discussion include:

• Enabling innovation by dialing back federal barriers to AI development and deployment
• Partnering with industry ton secure, standards-based export packages for AI models, software, and applications
• Expediting data center and semiconductor-fab buildouts through streamlined federal permitting
• Updating federal procurement so agencies buy only from frontier LLM providers that meet specified objectivity and safety criteria

Financial services teams are juggling overlapping obligations as state AI laws advance while federal policy continues to shift.  The result is real uncertainty about which rules drive exams and which controls must be proven first.  This session turns the patchwork into an action plan tied to concrete use cases in lending, underwriting, fraud, and customer operations. Topics of discussion include:

• Mapping state and federal obligations to specific financial use cases so teams scan prioritize what to build and validate first
• Translating examiner expectation into checklists for documentation, reporting, and evidence so audits move faster
• Harmonizing privacy, model risk, and consumer protection controls so one control set satisfies multiple regimes Stress-testing conflicts of law and jurisdiction so global programs do not break when products cross borders

Fraudsters now use synthetic identities, deepfakes, and automated mule networks.  In response, banks are piloting models that surface patterns traditional rules miss.  Examiners still expect the Bank Secrecy Act fundamentals while asking for proof that AI improves quality and does not create blind spots.  This session shows what “good” looks like, from model documentation to SAR workflow evidence and sanctions screening integration.  Topics of discussion include:

• Detecting fast-moving red flags with feature engineering and triage playbooks so alerts are more accurate and timely
• Documenting end-to-end SAR processes so automation raises quality and withstands subpoenas and exam review
• Integrating sanctions screening with entity resolution and watchlist logic so false positives drop without raising risk
• Operationalizing model monitoring so drift, bias, and escalation thresholds are visible to both compliance and auditors

Choosing between buy, build, or co-develop decisions is hard when model performance, cost, and explainability all matter, especially to the Board. This showcase compares real use cases that survived procurement, security, and legal review, including lessons learned from sandbox pilots.  Topics of discussion include:

• Selecting high-value use cases with measurable ROI, so early wins unlock budget and sponsorship
• Designing guardrails and acceptance criteria so pilots convert to production without rework
• Evaluating vendor claims with proof requirements and test data strategies so marketing does not outpace reality
• Crafting the board pitch with cost, risk, and control narratives so approvals land on time 

Regulators and law enforcement agencies are sharpening their focus on algorithmic decisioning, predictive analytics, disclosures, and cyber resilience around AI. Expect increasing scrutiny into documentation, policies and procedures, supervision, and AI disclosures. This session translates public statements and recent enforcement actions from the SEC, FINRA, and DOJ into governance frameworks, oversight, and regulatory defense. Topics of discussion include:

• DOJ AI priorities and prosecutions
• SEC’s Division of Enforcement AI priorities and enforcement actions
• SEC 2026 AI examination priorities
• FINRA’s statements on AI deployment, including GenAI and agentic AI
• Best practices for AI policies and procedures, governance, and disclosure risk management

Model risk leaders must show how AI fits inside SR 11-7 and OCC expectations, while preserving performance.  The exam room wants independent validation, challenger testing, and clear lines of accountability across the model lifecycle when models are used to make decisions.  But how does this look when AI is used to understand and implement the regulations themselves?

This session will make the practical case for incorporating Agentic AI in regulatory compliance processes, while also using it as a paradigm for understanding examination standards applicable to AI in connection with model risk management. Topics of discussion include:

• Application of AI in the context of regulatory change lifecyles
• Practicality of using Agentic AI with respect to changing AML and sanctions requirements
• Meeting regulatory expectations with respect to use of decision making models

Credit and pricing models face the toughest fairness questions, and new state laws heighten expectations for testing and transparency.  Data quality, feature selection, and outcome monitoring all factor into legal risk and examiner comfort.  This session shows how programs are proving fairness without sacrificing accuracy.  Topics of discussion include:

• Selecting features and mitigation techniques so fairness improves, while strength remains intact
• Generating clear explanations and adverse-action content so customer communications satisfy exam standards
• Validating data lineage and representativeness so drift and legacy bias do not erode outcomes over time
• Running pre-deployment and post-deployment fairness tests so corrective actions are timely and defensible

Strong AI programs combine policy, role clarity, and privacy-by-design with vendor and product controls.  Regulators are also probing consequential decisions in credit and insurance for transparency, access, and appeals.  This session delivers the playbook for making governance visible and auditable.  Topics of discussion include:

• Forming an AI risk committee and RACI so decision rights and accountability are visible across lines of defense
• Writing vendor clauses for IP indemnity, audit access, service levels, and bias thresholds so contracts carry control weight
• Tightening access, logging, and human-in-the-loop requirements so outcomes are reviewable and correctable
• Synchronizing privacy impact and AI impact assessments so one review satisfies multiple obligations