noworkleft.com

Home › Jobs › Private Detectives and Investigators

Protective Service · SOC 33-9021

Private Detectives and Investigators

Gather, analyze, compile, and report information regarding individuals or organizations to clients, or detect occurrences of unlawful acts or infractions of rules in private establishment.

Will AI replace this job?

The data cannot say, and anyone who gives you a percentage for “replaced” is guessing. What it does show is a gap between three things, for the 14 tasks O*NET lists for this occupation:

Could an LLM do it?

7%–21%

of tasks: an LLM alone could do the task or cut its time by half. Human raters say 21%, GPT-4 says 7%.

Counting tasks that need extra software around the model: 50%–79%.

Is it happening?

0 of 14 tasks

show measurable use in Claude conversations (Anthropic Economic Index). Anthropic’s occupation score, which counts only tasks that were also rated feasible, is 0%. Observed is a subset of “could” by construction.

Claude only, mostly US users, sampled periods in 2025. Not all AI use, and not employment.

What is left?

2 of 14 tasks

were rated “no LLM speed-up” by every rater (so, by Anthropic’s method, they cannot show up as observed use either). That is the part of the job these studies see no sign of AI touching.

It does not mean those tasks are safe. It means these methods found nothing there.

Task by task

Sorted with the tasks seen in use first. The three coloured squares are the three raters, in this order: human raters, GPT-4, GPT-4 with an alternative rubric. “Usage score” is Anthropic’s published task_penetration value; 0 means none seen.

LLM aloneneeds extra softwareno speed-up
Task (O*NET 31.0)Could an LLM?Claude usage score
Write reports or case summaries to document investigations. all raters none seen
Search computer databases, credit reports, public records, tax or legal filings, or other resources to locate persons or to compile information for investigations. all raters none seen
Expose fraudulent insurance claims or stolen funds. all raters none seen
Obtain and analyze information on suspects, crimes, or disturbances to solve cases, to identify criminal activity, or to gather information for court cases. all raters none seen
Conduct private investigations on a paid basis. 2 of 3 raters none seen
Conduct personal background investigations, such as pre-employment checks, to obtain information about an individual's character, financial status, or personal history. 2 of 3 raters none seen
Question persons to obtain evidence for cases of divorce, child custody, or missing persons or information about individuals' character or financial status. 2 of 3 raters none seen
Investigate companies' financial standings, or locate funds stolen by embezzlers, using accounting skills. (supplemental) 2 of 3 raters none seen
Alert appropriate personnel to suspects' locations. (supplemental) 2 of 3 raters none seen
Count cash and review transactions, sales checks, or register tapes to verify amounts or to identify shortages. (supplemental) 2 of 3 raters none seen
Observe and document activities of individuals to detect unlawful acts or to obtain evidence for cases, using binoculars and still or video cameras. 1 of 3 raters none seen
Confer with establishment officials, security departments, police, or postal officials to identify problems, provide information, or receive instructions. 1 of 3 raters none seen
Testify at hearings or court trials to present evidence. no rater none seen
Perform undercover operations, such as evaluating the performance or honesty of employees by posing as customers or employees. (supplemental) no rater none seen
How to read this. These are averages over task lists written for the US labour market, rated by people and by GPT-4 using early-2023 model capabilities, next to a count of what one company’s chatbot users actually ask for. A task being “exposed” says nothing about whether employers will cut jobs, shift work, or hire more of you. See how to read an exposure score.

Try it with your own task list

Sources: task list from O*NET® Database (U.S. Department of Labor, CC BY 4.0; modified by noworkleft.com); “could” labels from Eloundou et al., GPTs are GPTs (OpenAI/University of Pennsylvania, MIT licence, labels from early-2023 models); “seen in use” from the Anthropic Economic Index (Claude usage only, CC BY). Dates, versions and limits.

Other jobs in Protective Service

Data built 2026-11-06. O*NET 31.0. Task match rates: 99.4% with the OpenAI/Penn labels, 99.2% with Anthropic usage.