← All finance signals
Governance

The Always-On Audit

US regulators are weighing compliance AI from both sides: FinCEN would credit banks’ AI tools, while FINRA warns that agent reasoning can be hard to audit.

What is it?

Regulators count AI inside compliance: FinCEN would weigh a bank’s AI tools in supervisory decisions, and FINRA sees agents monitoring trading-abuse alerts at member firms. Whether continuous checking earns trust or just proves activity stays unproven.

Why now?

Supervisors are moving early on this technology: a FinCEN proposed rule of April 2026 names AI as a factor, while FINRA flags that multi-step agent reasoning can be hard to trace for audit. The tool is credited and questioned at once, and that is a different kind of governance.

Who’s building it?

RegTech vendors build the tools; banks and broker-dealers deploy them, while FinCEN and FINRA decide what the tools are worth.

RegTech vendorsBanks and broker-dealersTier-1 banksBoards & auditorsRegulators
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Builders — RegTech vendors and compliance-software makers building the mapping tools.
Buyers — banks under anti-money-laundering supervision and broker-dealers trialling AI agents.
Gatekeepers — FinCEN and FINRA, deciding how much an automated check counts.
What’s at stake
A supervisor that weighs a bank’s AI tools turns compliance tooling from a cost centre into evidence, a repositioning rather than a rebrand.
Vendors compete on speed of mapping new rules.
Manual review work is shrinking into a standing function.
What’s unsettled
Whether continuous monitoring actually reduces risk, or just documents it better.
Whether smaller institutions can afford tooling a supervisor may come to expect.
Whether regulators will accept automated mapping as compliance, or only as evidence of effort.
02 · How the demand matureda progression, not a timeline
Manual reading
A specialist reads regulatory text line by line, slow and costly.
Automated mapping
Software reads the same text and maps it to internal controls.
Continuous monitoring
Checks run standing rather than annually, catching drift the moment it appears.
Supervisory credit
A regulator weighs whether a bank’s AI tools show its programme works.
03 · Where the market goesthe forks, not a prediction
→Consolidation or fragmentation — decided by whether continuous compliance becomes a shared utility banks buy together, or a proprietary edge each institution builds alone.
→Regulators may formalise real-time mapping as the expected standard, not a competitive choice.
→Smaller institutions may be priced out, leaving compliance automation concentrated among a handful of vendors.
→Trust may come to depend on the audit trail, not the outcome itself.
Everything here turns on trust: whether continuous, machine-read monitoring is treated as real assurance, or as a faster way to produce the paperwork regulators already expected.
04 · The Compliance-as-Feature deskadjacent markets, moving now
Regulated Reality
Election deepfakes now carry legal consequences.
30 US states now have election-deepfake laws.
Public Citizen
The Explainability Demand
EU law now mandates labelling AI-generated content.
EU AI Act Article 50 mandates AI-content transparency labelling.
European Commission
Certified Before Release
Independent labs now test AI models like crash tests.
A crash test for AI.
CNN Business
05 · The Wire
01
FinCEN proposal weighs banks’ use of AI tools
U.S. Financial Crimes Enforcement Network ↗
02
FINRA sees AI agents moving into early use at member firms
FINRA ↗
The panel on this signal
They agreed on
–Compliance is shifting from an annual audit exercise into a standing, continuously monitored responsibility that supervisors now weigh directly.
–Supervisors are crediting AI tools whose reasoning can be hard to audit.
–The work genuinely moves from specialist manual reading into automated, real-time mapping.
The regulatory analyst disagreed on
–Continuous monitoring systems generate a more complete audit record, which is distinct from demonstrating an actual reduction in underlying compliance risk; the two outcomes are not necessarily the same and current reporting has not separated them.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.
✦Signal changelogre-read as evidence changed
2025
Boards are told 2026 is the year AI-driven compliance becomes a measure of institutional trust, with most buyers planning new tools.
→
2026
Regulators now name AI tooling in supervision: FinCEN’s April 2026 proposed rule lists a bank’s use of innovative tools such as AI among factors. U.S. Financial Crimes Enforcement Network ↗