Why the camera room still needs a watcher
Ask whether AI will replace gambling surveillance officers and you have to look at what the shift actually involves. The work is watching, judging and documenting. Officers monitor the gaming floor and recorded video for cheating, theft, card marking, chip walking and underage play. Software is good at the watching part. It is far weaker at the second part: deciding that a flagged clip is a scam, naming who did what, and standing behind that call when someone pushes back.
The paperwork carries weight too. Surveillance logs, incident reports and evidence packages go to property management, to police and to state gaming regulators. Officers interview employees and patrons, pull footage for licensing reviews, and may be asked to explain their findings under oath. A regulator wants an accountable, licensed person who saw the event and can be questioned about it. A model output with no name on it does not fill that role.
This is a small occupation. About 9,520 people held these jobs at a median wage of $43,370 (BLS, 2025), and BLS projects employment changing by -1.8% between 2025 and 2035. That is the honest shape of the pressure here: not the trade vanishing, but fewer monitor positions per property as analytics cover more cameras, and fewer easy entry points for someone starting out.
What software handles, what it assists, and what stays with people
The split at the top of this page sorts the duties three ways. Roughly 0% of task time falls into work AI can already carry: continuous camera coverage, motion and face matching, pulling clips by timestamp, and flagging bet or chip patterns that sit outside the norm. Machines do not get bored at 4 a.m., which is exactly where human attention fails.
A second slice, about 54% of task time, is assisted rather than handed over. Drafting a log entry, summarizing hours of footage into a short timeline, and cross-checking player accounts against anti-money-laundering indicators all go faster with a tool, then still need a person to check and sign. Our coverage figure for this job, the share of task time AI could handle today, reads 27 out of 100 on the coverage scale.
That leaves 46% of task time in the group the task list marks as needing a person. Interviewing a dealer, confronting a suspected cheat with a floor supervisor, briefing a regulator and testifying all sit there. So does the ordinary judgment of deciding which flags are worth escalating on a busy Saturday night.
How strong is the evidence on these tasks?
Thin, and we say so. The evidence grade for this occupation is D, our label for work where no study has tested AI against a qualified person doing this job. Because of that, we publish no parity number here. Vendor claims about detection rates are not the same as a measured head-to-head on real casino incidents.
What would settle it is specific: a blind test where video analytics and licensed surveillance staff review the same set of floor incidents, scored on detected events, false flags and whether the resulting report held up with a gaming regulator. Audited AML case work would help too. Until something like that is published, the evidence list on this page stays short, and our grade stays where it is. The method behind the scores explains how grades move when a study lands.
When this could change
Most likely between 2041 and 2055 (8 in 10 of our scenarios). The replacement-year method explains what that window does and does not claim.
Two things could pull it earlier. First, there is no robot hardware problem to solve: the robotics tier for this job is none needed, because the tools are cameras, servers and screens that properties already own. Second, the cost comparison on this page is lopsided. Software licensing sits well under the cost of staffing a monitor desk around the clock, and that gap is what shrinks headcount.
Two things hold it back. Gaming regulation ties surveillance duties to licensed, accountable people, and rules change slowly state by state. And the end product is often evidence: an interview, a statement, a sworn account. A model can prepare that package; it cannot be the witness.
What to do: get named on the investigation and compliance side of the room, not just the monitoring side.
How to stay needed in surveillance and investigations
Lean into the duties the task list leaves with people. Run the interviews and statements, so you are the one who turns a flagged clip into a usable case file. Own the regulator relationship, including license reviews, incident notifications and audit requests. And take the responsible-gambling and exclusion calls, where a wrong flag has real consequences for a patron.
Two skills pay for themselves. One is anti-money-laundering and fraud casework: suspicious activity reporting, structuring patterns, and how a case is built for compliance review. The other is running the analytics stack itself, including tuning alerts, cutting false positives and writing clear findings a manager or a court can follow.
If you are weighing a move, the closest work sits nearby: security guards, private detectives and investigators and retail loss prevention specialists. You can see the rest of the group on the other protective service workers family page, or read the wider picture for the arts and entertainment sector.
Two more useful stops: put this job next to one of those options on our side-by-side comparison, and see how the major assistants answer the same question in what the AIs say.