Police and Detective Supervisor

This page's structural exposure score is 49 out of 100, in the Moderate Risk band, and a lower score means more exposed. The two sources behind it disagree: OpenAI's task ratings read this occupation as more exposed than average, while the AIOE index reads it as somewhat less exposed. Either way it is a score about tasks on paper, and it says nothing about who is legally allowed to perform them.

Under this site's research checklist, a score below 60 needs a documented legal or physical wall before we call a job protected in practice. On whether AI will replace these supervisors, the evidence below says not on current evidence, with medium confidence.

The first wall is legal. Supervisors in this occupation are normally sworn officers promoted from the ranks: BLS says promotions to corporal, sergeant, lieutenant and captain usually follow a written examination and on-the-job performance. Two of the tasks O*NET lists, ordering the detention of witnesses and suspects and conducting raids, are exercises of police authority. In California, Penal Code section 830.

1 names police officers, chiefs of police and sheriffs as peace officers with authority to act on public offenses; it does not list ranks such as sergeant separately. We read only California's statute, so we do not claim every state works the same way. The statute describes people employed in a role, not software, and O*NET reports that 59% of respondents say an error in this job has an extremely serious consequence.

The second wall is physical presence. O*NET's work-context data show that 96% of respondents hold face-to-face discussions every day, 62% work outdoors exposed to all weather every day, and 47% deal with unpleasant or angry people every day. A supervisor who arrives at a scene, steadies a squad and answers for what happened is doing something a drafting tool does not.

AI is already inside the paperwork half of the job, and that matters. Axon's Draft One writes a first report narrative from body-camera footage, and KJZZ reported in August 2025 that Arizona departments including Phoenix, Gilbert, Scottsdale and Tempe use it.

California Penal Code section 13663, effective January 1, 2026, requires agencies to have a policy that official reports written fully or partly with AI name the AI program, carry an AI disclosure statement and bear the signature of the officer who prepared the report, verifying they reviewed it. The agency must also keep the first AI draft and an audit trail. That adds a review duty for the signing officer rather than removing one.

In the same KJZZ interview, Mother Jones reporter Tekendra Parmar said Fort Collins, Colorado, and Lafayette, Indiana, had turned off the ability to check whether a report was AI-generated, so oversight varies by agency.

Body-camera review is the other live front. Truleo's software transcribes footage, flags language such as insults or threats as pending, and sends those items to supervisors, who can add a follow-up label or remove the flag (The Regulatory Review, September 3, 2026, summarizing a study by Vanderbilt's Farhang Heydari and coauthors). The tool filters what a sergeant looks at; a person still decides.

We searched for a department that cut supervisor positions because of AI and found none. That is a statement about our search, not proof it has not happened. The exposed tasks are real: scheduling, report review, policy drafting, budgets and correspondence. A department could plausibly need fewer supervisors per squad if those shrink, but no source we read shows it. Microsoft's Copilot study gives this occupation an AI applicability of 0. 083, below the 0.

159 average across 785 occupations, which is why the real-world usage band reads Medium.

BLS Table 1.2 projects 3.3% growth from 2025 to 2035, 160,200 to 165,400 jobs, with about 10,600 openings a year. For the wider police and detectives group, BLS says demand varies by location, driven largely by local and state budgets. Budgets, not software, are the main swing factor we found. BLS OEWS (May 2025) puts median pay at $106,040, with the 10th to 90th percentile range at $63,350 to $165,110.