EXPOSED
The bulk of the job — reading regulations, screening transactions and filings against rule sets, drafting policies, assembling audit evidence, writing findings reports — is exactly the text-and-pattern work current AI does at usable quality, and RegTech vendors are already shipping it. What holds is the accountable layer: the named Chief Compliance Officer or BSA Officer whom regulators can personally sanction, the judgment call on whether an ambiguous transaction is escalated, and the human who sits across from an examiner. The modal compliance officer today sits closer to the screening-and-documenting tier than the accountable tier, which is why this scores as exposed rather than safe.
Headcount grew steadily across the period.
Median pay $69,050 → $80,730 -6.5% in real terms
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing. The marked year is 2020.
BLS projection, 2024–2034
+3% 418,000 → 430,300 on the projections basis
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +3% more of these jobs by 2034. Demand for the output is growing faster than the work is being automated away — the mechanism BLS gives for software developers, and the combination people most often misread as an error.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~33,300 openings a year on average, including replacing people who leave.
CoronerExaminerExporterImporterCargo AgentCargo BrokerEntry WriterExport AgentImport AgentCustoms AgentTest ExaminerAction OfficerBoarding AgentCounty CoronerCustoms BrokerField ExaminerForms ExaminerPassport ClerkCity SanitarianCustoms AnalystDriver ExaminerState InspectorAir Export AgentAir Import Agent
Holding it up: judgment & accountability . Weakest point: embodiment .
Mixed — a routine tier and a judgment tier Alert adjudication, control testing, policy mapping to new rules, KYC file review and periodic risk assessments are all document-in/document-out work that transaction monitoring and RegTech platforms already draft or clear at first pass — a 7 rather than a 3 because interviewing a business line about why a control failed, and negotiating remediation deadlines with people who don't report to you, still needs a human in the room.
Some physical or field component There is a physical tail — walking the trading floor or branch, observing a manufacturing line for OSHA or EPA conformance, pulling paper files during an on-site exam, and travelling to subsidiary offices — but it is scheduled, indoor, and a minority of the calendar, which puts it just above pure desk work rather than in the field-inspector range.
Certification preferred, not legally required No licence is required to be a compliance officer, and most of the 417,000 in this SOC hold CAMS, CRCM or CCEP credentials that are employer preferences rather than a legal gate — the 9 reflects that a specific subset (the designated BSA/AML officer under 31 CFR 1020.210, the FINRA-registered Series 14 supervisor, the named CCO) can be individually fined or barred, and that personal exposure is real but attaches to a few titles, not the job family.
Meaningful discretion Deciding whether an ambiguous wire is escalated to a SAR, whether a conflict is disclosable or disqualifying, and whether to self-report to a regulator are calls made on incomplete facts with legal consequences either way and no procedure that closes them — the 13 rather than 16 reflects that in most shops those calls are ratified by the CCO, legal, or a committee before they leave the building.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (7/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (9/20) is whether the law requires a licensed human to sign. Trust premium (8/20) is whether buyers specifically pay for a person. Judgment and accountability (13/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 30 of this occupation's 42 points (71%).
Embodiment (5/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
Emergency Management Directors EXPOSED
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 58/100, still EXPOSED.
Task-mix shift: if vendor tooling absorbs alert triage, evidence assembly and policy drafting, the residual role is escalation decisions, SAR/no-SAR calls, examiner defense, and adjudicating where the model was wrong. The occupation genuinely has two tiers, and the surviving tier owns consequential ambiguous calls.
Model-risk-management rules (SR 11-7 extended explicitly to AI compliance tooling, or NYDFS Part 504-style transaction-monitoring certification broadened) requiring a named human to validate and sign off on each automated screening model before deployment and after each material change. NYDFS Part 504 already requires an annual certification by a senior officer for AML transaction monitoring systems — extending that pattern to other jurisdictions and to sanctions/fraud models is a watchable event.
Explicit human-in-the-loop mandates for adverse automated decisions: EU AI Act Article 14 human oversight duties for high-risk systems (including creditworthiness and employment screening) and Colorado SB 24-205-style state AI acts, where a compliance function must own and document the override decision on individual adverse outcomes.
Named-individual accountability regimes extending below the CCO level: FinCEN/FFIEC-style requirements that a designated BSA Officer personally attest to the adequacy of AI-driven alert triage, or a US analogue of the UK FCA/PRA Senior Managers & Certification Regime that assigns statements of responsibility to compliance staff below the C-suite. Also concrete: SEC Rule 206(4)-7 annual review being interpreted to require a named human's signed certification that model-generated surveillance output was reviewed, following the Gurbir Grewal-era 'compliance officer as gatekeeper' enforcement posture.
Same two-tier compression: the remaining work becomes contested interpretation of new rules with no precedent, negotiating remediation scope with an examiner, and internal investigations involving interviews and uncooperative witnesses — none of which current tools do at usable quality. Note this raises the score of the remaining jobs, not the number of them.
The limit. Trust premium has no realistic route here — the buyer is an internal budget holder treating compliance as cost, not a client paying for a human; only regulators create the demand. Embodiment has none. And every liability lever concentrates protection in a shrinking named-officer layer while the screening tier is displaced, so headcount can fall sharply even as the register score for surviving roles rises.
| New York-Newark-Jersey City, NY-NJ | 23,380 | $101,050 +25% |
| Los Angeles-Long Beach-Anaheim, CA | 17,700 | $94,810 +17% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 13,820 | $100,200 +24% |
| Dallas-Fort Worth-Arlington, TX | 12,530 | $77,760 -4% |
| Boston-Cambridge-Newton, MA-NH | 10,920 | $103,940 +29% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 10,530 | $79,770 -1% |
| Atlanta-Sandy Springs-Roswell, GA | 9,170 | $72,810 -10% |
| Houston-Pasadena-The Woodlands, TX | 8,450 | $78,860 -2% |
| San Jose-Sunnyvale-Santa Clara, CA | 2,690 | $123,730 +53% |
| San Francisco-Oakland-Fremont, CA | 7,500 | $118,890 +47% |
| Bellingham, WA | 640 | $109,550 +36% |
Goldman Sachs · MVB Bank · IAG
FF News reports MVB Bank has partnered with Bretton AI to automate back-office operations supporting its fintech banking business.
Insurance Business reports that insurer IAG is adopting agentic AI in areas where conduct risk is highest, according to company statements.
JD Supra reports that Connecticut has enacted an AI law imposing compliance obligations on employers, including around AI use in employment decisions.
A legal blog reports Connecticut has enacted a new AI law with compliance obligations for employers, including in employment decision-making contexts.
Norton Rose Fulbright reports Colorado has enacted a revised version of its artificial intelligence law governing AI systems, including those used in employment decisions.
A law firm reports that the Illinois Department of Human Rights has issued regulations governing employers' use of AI in employment decisions.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.