ChatGPT for HR Managers: 7 Prompts and the Rules to Check First

Drafting is where ChatGPT earns its place in HR: first versions of job ads, policies, manager scripts and onboarding checklists. Selection decisions are where it creates legal risk. SHRM's 2026 research found that 39% of organizations have implemented AI in HR, and 87% of HR professionals reported improved efficiency (respondent-reported). This article gives seven copy-paste prompts, current plan prices as of 16 October 2026, and the data and legal checks to run first.

This article does not promise a number of hours saved per week. Your savings depend on how many documents you draft and how much editing the output needs.

Is ChatGPT actually saving HR teams time?

Many HR professionals say yes, but the evidence is self-reported. In SHRM's research, 87% of HR professionals reported improved efficiency, 75% improved work quality and 70% increased creativity. These are survey answers, not measured time studies.

The same research shows how uneven the picture is. Where AI is used in HR, it is concentrated in recruiting (27%), HR technology (21%) and learning and development (17%). 56% of HR functions do not formally measure the success of their AI investments, and only 16% use ROI as a metric. The research draws on 1,908 HR professionals and 138 distinct HR tasks.

A practical takeaway: time a task before and after, for two weeks. If you cannot show the saving, you cannot defend the tool to finance or legal.

Which ChatGPT plan should an HR team use?

For anything touching employee information, a business plan with admin controls is the safer starting point. As of 16 October 2026, OpenAI's help center lists these prices:

A Business workspace requires at least two paid seats. Prices are in USD and OpenAI notes they may vary by country and currency. Enterprise pricing is not listed on those pages, so ask OpenAI's sales team.

On a personal account, turning off "Improve the model for everyone" means new conversations are not used to train OpenAI models. The path is Settings, then Data controls. These are OpenAI's own statements about its products, so have your IT or legal team review the terms before you rely on them.

What should you never paste into ChatGPT?

Never paste anything that identifies a person or reveals a confidential case, unless your company has approved the specific plan and use. That covers names, employee IDs, medical or leave details, pay data, investigation notes and candidate contact details.

A simple habit works well: replace real names with roles ("Employee A", "the hiring manager") and strip dates and locations that could identify someone. Then paste. This is a house rule, not a legal standard, and your data protection policy may be stricter.

Two checks before you start:

1. Which account are you on, personal or company workspace?
2. Does your employment contract, client contract or privacy policy restrict sending this data to a third-party processor?

If you run an agency serving EU or US clients, the second question matters more, because client contracts often limit where candidate data can go.

Which seven ChatGPT prompts do HR managers use most?

These are original templates, not tested benchmarks. Replace the bracketed parts, and edit every output before it leaves your desk. Start each session with a role line so the tone stays consistent: "You are an experienced HR manager at a [size]-person [industry] company in [country]. Write plainly. Flag anything that needs legal review."

1. Job description first draft

Draft a job description for [role title] reporting to [manager title].
Include: a 3-sentence summary, 6 responsibilities, 5 must-have
requirements, 3 nice-to-haves, and our pay range of [range] in
[currency]. Use inclusive language, no jargon, and no degree
requirement unless I state one: [degree rule]. Do not invent
benefits. Benefits are: [list].

2. Structured interview questions

Write 8 structured interview questions for [role], each tied to one
competency from this list: [competencies]. For each, add a
1-to-5 scoring guide describing what a weak, average and strong
answer looks like. Keep every question job-related.

3. Plain-language policy draft

Draft a [policy name] policy for a [size]-person company in [state or
country]. Use short sentences, an eighth-grade reading level, and
headings for: purpose, scope, rules, manager responsibilities,
exceptions, and who to contact. List any points that need
employment-lawyer review at the end.

4. Manager conversation script

I am coaching a manager who must tell an employee that [situation,
anonymised]. Write a 5-minute script with an opening, the facts, the
expected change, the support offered, and 3 likely employee
reactions with suggested replies. Keep the tone calm and factual.

5. Survey comment themes

Below are anonymised open-text survey comments with names removed.
Group them into no more than 6 themes. For each theme give a
one-line description, 2 representative comments, and one action
a manager could take. Do not guess who wrote any comment.
[paste comments]

6. Onboarding checklist

Create a 30-60-90 day onboarding plan for a new [role] in [team].
Show it as a table with columns: week, goal, owner, and check-in
question. Include day-one logistics and the first-week intros.

7. Wording cleanup for performance feedback

Rewrite this feedback so it is specific, behaviour-based and free of
personality labels. Keep the facts. Remove anything that refers to
age, gender, health, family status or other protected traits.
[paste anonymised draft]

For process-heavy tasks such as new-hire paperwork, a prompt only drafts the words. Our post on onboarding automation covers where software does the routing and e-signing. If you are mapping your wider tools, see the HR tech stack for a 100-person startup.

What makes a good HR prompt?

Four ingredients make the difference: a role, the context, the format and a limit. Prompts 1 to 7 all state who the reader is, what the output should contain, how it should be shaped and what the model must not do.

Add two habits on top. First, ask the model to list assumptions and anything it is unsure about, then check those yourself. Second, never accept legal statements, pay benchmarks or law citations at face value. Verify them against the statute, the regulator's site or your counsel.

Which uses of ChatGPT in HR carry legal risk?

Anything where the output influences a decision about a person is the high-risk zone. Drafting a job ad or a policy outline carries far less risk than ranking applicants or summarising a worker's performance for a termination decision.

A University of Washington study found three large language models favored white-associated names 85% of the time and female-associated names 11% of the time when ranking 550 real-world resumes. Note the limits: the models studied were open-source embedding models from Salesforce, Contextual AI and Mistral, not ChatGPT. The study shows the kind of risk, not a result for ChatGPT.

The rules below are a summary, not legal advice. Check the current text with counsel.

On the Colorado timeline, a federal magistrate judge stayed enforcement of the 2024 law on 27 April 2026 before the replacement was signed. On the EU side, the Digital Omnibus on AI was published in the Official Journal on 24 July 2026. Law firm Ogletree noted that employers may still want to prepare for the original deadline given the scope of the requirements.

For the wider compliance picture, see HR compliance software and what skipping it costs. If you are weighing AI interview tools rather than a chat assistant, read Paradox vs HireVue. For sourcing searches, the Boolean strings post covers methods that do not rely on a language model.

How do you set a simple AI rule for your HR team?

Write a one-page rule before the team starts. SHRM's research found 49% of organizations have policies regulating employee AI use, so many do not yet, and 57% of HR professionals in states with workforce-related AI regulations are not aware of those rules.

A workable one-page rule:

1. Approved accounts only. Name the plan and workspace. No personal accounts for work data.
2. Data rules. List what must be anonymised or never pasted (see the section above).
3. Allowed uses. Drafting, summarising anonymised text, formatting, brainstorming.
4. Restricted uses. Screening, ranking or scoring people needs legal sign-off first.
5. Human review. A named person checks every output before it is sent or filed.
6. Record keeping. Save the prompt and final version for anything that becomes a policy or candidate-facing message.
7. Review date. Revisit every quarter, because plans and laws are changing quickly.

What's still unclear?

Time saved. Weekly hours saved by HR prompts have not been measured in the sources cited here. The 87% efficiency figure is respondent-reported.

Colorado details. Seyfarth and Ogletree give 14 May 2026 as the signing date for SB 26-189 and 1 January 2027 as the effective date. Implementation details are left to future rulemaking.

Illinois rules. The Illinois summary used here dates from September 2024. Check the current statute and any Illinois Department of Human Rights rules before relying on it.

EU timing. The delay is in the published regulation, but employers with EU staff may still choose to prepare early.

Product terms change. OpenAI's plan names, seat types and limits have changed several times. Confirm prices and data terms on OpenAI's pages before you buy. Last checked: 16 October 2026.

Weakest point of this article. The prompts are templates written for this post and were not benchmarked. Quality will vary by model version, plan and how much context you supply.