An Event With
Executive Roundtable: Can You Defend Your Surveillance Programme?
Date: July 9, 2026
Time: 8:00 – 10:00 AM EDT
Location: PwC, 300 Madison Ave, New York, NY
About This Roundtable
Financial firms are under pressure to expand into new markets, desks, and communication channels without degrading control maturity. The problem is rarely detection alone. It’s the operating model that connects obligations, surveillance signals, investigations, and record-keeping into a single, regulator-ready control story.
This executive roundtable — hosted by Behavox — is a working discussion for senior compliance, surveillance, and technology leaders focused on the seams: where handoffs break, where evidence gets reconstructed too late, and how leading firms are building auditable workflows that hold up under examination.
The stakes are rising. Agentic AI is already embedded in surveillance workflows — flagging alerts, drafting investigation summaries, escalating cases — before governance frameworks exist to govern it. SR 26-2, the Fed, OCC, and FDIC’s updated model risk guidance, explicitly excludes generative and agentic AI from its scope. A June 2026 survey found nearly three in four banks cannot confirm they can shut down a malfunctioning model or report a failure to regulators. The tools are ahead of the frameworks — and regulators are starting to notice.
2026 is less about reacting to new rules and more about proving existing obligations are embedded, scalable, and defensible. There is a difference between a program that survives an exam and one that signals institutional maturity.
Discussion Topics
Agenda
| Time | Overview |
| 8:00 AM EDT | Arrival and Networking |
| 8:30 AM EDT | Roundtable Discussion |
| 10:00 AM EDT | Closing Remarks |
A Special Guest Moderator
There will be a special guest moderator for this roundtable: a recognised Compliance leader with extensive experience across trading and compliance risk. We look forward to introducing them on the day.
Chatham House Rule applies, ensuring open and confidential discussions.
This invitation does not constitute gifts/entertainment and there is no entrance fee and no preparation required. Should you have additional questions or are unable to register online, please reply to [email protected] and a member of our team will be happy to assist you.
Trade and Comms Surveillance Were Never Designed to Talk to Each Other — Now Regulators Expect Them To
Cross-asset conduct cases require connecting a trade event to a communication to a policy exception to a prior alert. Most firms have three separate teams, three separate queues, and no unified case view. FINRA’s 2026 report explicitly calls for enhanced cross-product and cross-customer manipulation surveillance. This discussion examines the organizational, data, and technology challenges of building a converged program.
Agentic AI in the Surveillance Workflow: Accountability, Handoffs, and the Governance Gap
Agentic AI is live in compliance workflows — flagging alerts, drafting investigation summaries, escalating cases without human prompting — yet most firms have no formal governance framework for the systems they’ve already deployed. SR 26-2 explicitly excludes the systems most aggressively deployed in compliance. This discussion examines what rigorous oversight requires and who owns accountability when an agentic system makes a wrong call.
From Exam-Ready to Exam-Proof: Building a Surveillance Program Regulators Trust
A converged surveillance program and a governed AI layer are necessary — but neither is sufficient unless you can reconstruct a defensible record under examination. Documentation practices, investigation workflow design, and escalation chains matter as much as the underlying technology. This discussion explores what “exam-proof” actually looks like and the gap most firms have between their control design on paper and their ability to prove it holds up when something surfaces.
