Why Autonomous Agents Are the Next Major Risk in Finance - Heather Domin

On this episode of the FINOS Open Source in Finance Podcast, host Grizz Griswold speaks with Heather Domin, Global Head of the Office of Responsible AI and Governance at HCLTech. Heather shares her 20-year journey in data science and AI ethics, discussing why 76% of organizations delay AI deployments over responsible AI concerns. She explores the evolution from simple model risk to autonomous compounding risks in agentic workflows, the rise of the "Agentic Web" across financial transactions, and how the FINOS AI Governance Framework (AIGF) enables regulated banks to deploy audit-ready AI guardrails without hindering innovation velocity. 🕒 Timestamps:0:00 Welcome to the FINOS Open Source in Finance Podcast0:34 Executive Women’s Forum & Speaker Intro1:18 OSFF New York Announcement & Fluxnova Con Preview2:43 Heather Domin’s Journey: Data Science to Responsible AI Lead4:16 The Evolution of AI Ethics: From Research to the EU AI Act7:12 Why 76% of Organizations Delay AI Over Governance Concerns8:51 Evolving Risks: From GenAI Hallucinations to Autonomous Agentic Chains10:45 Managing High-Risk Financial AI vs. Retail Use Cases12:12 Overcoming AI Fear: Policy, Process & Technical Controls16:28 Reasons for Optimism: Upskilling & Inbuilt Safety Controls18:49 The Next Wave: Financial Transactions on the "Agentic Web"21:22 OSFF Talk Preview: Operationalizing the FINOS AI Governance Framework (AIGF) 📊 The Problem: The AI Deployment Bottleneck & Compounding Agent RisksThe 76% Deployment Freeze: HCLTech research reveals that 76% of enterprises delay AI rollouts due to responsible AI, governance, and regulatory uncertainty. Evolving Risk Profiles: Moving from generative AI text outputs to autonomous agentic chains introduces compounding, multi-step transaction risks that traditional model risk management frameworks cannot capture. Regulated Capital Risk: Financial institutions handle high-risk customer assets where un-governed agentic decisions carry severe legal, reputational, and financial consequences. 🏗️ The Solution: Responsible AI Governance & The FINOS AIGFHeather Domin details HCLTech's approach to operationalizing responsible AI: Control by Design: Embedding policy rules, threshold limits, and entitlement checks directly into AI platforms so developers innovate inside safe, automated guardrails. FINOS AI Governance Framework (AIGF): Leveraging open-source, consensus-driven control definitions tailored specifically for financial services to standardize risk mitigations across the software lifecycle. Governing the Agentic Web: Preparing enterprise architectures for cross-firm autonomous agent interactions, ensuring multi-step financial transactions across web protocols maintain auditability and human oversight. ⚙️ Why This Matters for Financial EngineeringUnlocking Scaled Production: Transitions financial institutions past prototype paralysis by establishing verifiable guardrails that satisfy regulatory mandates (such as the EU AI Act) without slowing engineering velocity. Human-in-the-Loop Orchestration: Ensures high-impact financial actions executed by autonomous agents retain explicit human oversight, verification checkpoints, and incident logging. 🌐 More about FINOS: https://www.finos.org/ 📧 Join our newsletter: https://www.finos.org/sign-up 🎙️ Listen to our Open Source in Finance Podcast: https://www.youtube.com/@FINOS/podcasts LinkedIn: https://www.linkedin.com/company/finosfoundation #FINOS #OSFFNewYork #AIGovernance