I am a police officer looking for practical, policy-conscious AI prompts for routine work. Create a copy-paste prompt pack for an officer, supervisor, or investigator. The prompts must support professional judgment and documentation; they must not replace agency policy, legal training, probable-cause analysis, dispatch procedures, evidence handling, or supervisor review. My context: [ROLE AND ASSIGNMENT] [AGENCY POLICY LIMITS] [COMMON TASKS]
[STATE OR LOCAL CONTEXT] [TOOLS AND AUDIENCE] Produce 10 ready-to-copy prompts, each with 2 to 5 bracketed inputs and a one-sentence note on safe use. Include prompts for: organizing case or incident facts before writing; converting notes into a neutral report outline; preparing a shift briefing; creating a training scenario; summarizing a policy document supplied by me; drafting a plain-language victim or community update; preparing interview-question topics from known facts; creating a follow-up checklist; reviewing a report for ambiguity or unsupported conclusions; and preparing a supervisor handoff. Every prompt must instruct the AI to preserve uncertainty, distinguish observed facts from witness statements and inferences, use neutral non-inflammatory wording, omit protected or unnecessary personal information, and flag information requiring policy, legal, or supervisory review. For report work, require the model to identify gaps rather than fill them. Do not generate arrest advice, tactical instructions, surveillance plans, use-of-force guidance, legal conclusions, personal data, or fabricated narrative details. End with a short “before you use AI” checklist covering approved agency tools, data entry restrictions, records retention, disclosure obligations, and human verification. Before answering, check that each prompt avoids asking the model to make a legal determination or invent a fact. Ask up to 3 clarifying questions only if a required input is missing.
Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [ROLE AND ASSIGNMENT] | Enter your role, unit, shift, and the kinds of calls or cases you commonly handle. | Patrol officer on evening shift in a mid-size city; frequent property crimes, traffic collisions, disturbance calls, and victim follow-up. |
| [AGENCY POLICY LIMITS] | Describe approved AI use, report-writing policy, confidentiality limits, and required review procedures. | No names, addresses, case numbers, body-camera content, or CJIS data in public AI tools; all reports must be written and approved in the RMS. |
| [COMMON TASKS] | List the recurring documentation, briefing, training, or communication tasks you want help with. | Shift briefings, training scenarios, report self-review, victim updates, and organizing follow-up items. |
| [STATE OR LOCAL CONTEXT] | Enter jurisdictional context that affects terminology or approved public communication. | Ohio municipal agency; public communication must go through the public-information office when it concerns an active case. |
| [TOOLS AND AUDIENCE] | State which approved systems you use and who will read the resulting material. | Approved internal assistant only; outputs are for the officer, sergeant, or training coordinator. |
How to use
- Enter only information permitted by your agency’s policy and approved tool rules.
- Copy the prompt that matches the task and replace its bracketed fields with de-identified facts.
- Compare the output with your notes, policy, and records before moving it into any official system.
- Follow up with: “Review this draft only for ambiguity, missing factual support, and privacy risks; do not rewrite or add facts.”
Variations
Report outline
Use this to organize notes before writing an official report.
Using only these de-identified notes, [FACT NOTES], create a neutral incident-report outline for [INCIDENT TYPE]. Separate officer observations, statements attributed to each source, actions taken, collected evidence, and unresolved facts. Preserve time sequence and flag contradictions or missing details. Do not supply names, legal conclusions, probable cause, or details not present in the notes. This is an organization aid only; I will write and verify the official report under [AGENCY REPORT POLICY].
Training scenario
Use this to create a discussion-based training exercise.
Create a discussion-based training scenario for [AUDIENCE] on [TOPIC], using [AGENCY POLICY EXCERPT] and [LEARNING OBJECTIVE]. Provide a short fictional setup, staged decision points, facilitator questions, expected policy references, and a debrief. Do not give tactical, use-of-force, surveillance, or legal advice beyond the policy excerpt. Mark every item that needs review by a training officer, legal adviser, or command staff before use.
Victim update
Use this for a plain-language draft that is subject to approval.
Draft a respectful victim-status update using [APPROVED CASE FACTS], [CURRENT STATUS], [NEXT CONTACT OR RESOURCE], and [COMMUNICATION POLICY]. Write 120 to 160 words in plain language, state only verified information, avoid promising an outcome or timeline, and avoid sensitive investigative details. Include a sentence directing urgent questions to [CONTACT METHOD]. Flag anything that requires investigator or public-information approval before sending.
Tips
- Treat an AI draft as unverified working material until you compare it line by line with notes, recordings, reports, and policy.
- Do not paste personally identifying, criminal-history, dispatch, victim, or evidence information into a tool unless your agency explicitly approves that data use.
- Ask the tool to surface gaps and ambiguities rather than “improve” a narrative; this reduces the risk of accidental additions.
- Keep the original notes and disclose AI use when required by agency rules, prosecution policy, discovery rules, or records-retention requirements.
FAQ
Can I use AI to write an incident report?
Use it only if agency policy permits it and only with approved systems and data. The officer remains responsible for factual accuracy, legal sufficiency, disclosure obligations, and the final report.
Can AI tell me whether I have probable cause?
No. Probable cause and other legal judgments require your training, current law, agency policy, and appropriate supervisory or legal review.
What information should never go into a public AI chatbot?
At minimum, avoid names, addresses, case numbers, criminal-justice information, protected victim information, evidence details, credentials, and any material prohibited by agency policy.