ChatGPT for HR: Drafts, Policies, and Human Review

ChatGPT is useful for HR teams in the same way it is useful for any team that produces a lot of structured written communication: drafting, reviewing, and improving documents against a set of criteria. HR teams use it to write job postings, review policies for plain-language clarity, prepare onboarding materials, and draft templated correspondence. The key discipline is keeping confidential employee information away from the tool.

The decisions an HR team makes about how to use ChatGPT, what it is not allowed to touch, and who must review its output before anything reaches an employee are the most important part of the workflow. Tools like fenbs keep those approved rules and the reusable prompts that go with them in one shared place, so a new HR team member starts from the same agreed standards rather than working out the boundaries from scratch.

What HR teams actually use ChatGPT for

The tasks that HR teams find most useful fall into four categories: creating and improving written documents, preparing for structured conversations, researching current practices and legal requirements, and checking existing materials for consistency or gaps. Within each category, ChatGPT works best when the HR professional provides the context, checks the output, and owns the result.

ChatGPT does not access your HR system, read your employee files, or know your organization's specific policies unless you tell it. Every session starts fresh unless you are using a paid plan with memory or a custom setup. That is a limitation, but it is also a reminder that the tool is an assistant you direct, not an autonomous agent making HR decisions.

Writing job descriptions and postings

Job description drafting is one of the most common HR uses for ChatGPT. Give it the role title, the key responsibilities, the required qualifications, and any specific requirements for your organization or location, and it will produce a workable first draft. The draft will need editing: ChatGPT tends toward generic language, and a good job description should reflect your actual culture and environment.

The review step matters here for compliance as well as quality. Phrases that seem neutral can constitute age discrimination or suggest an expectation that conflicts with the Americans with Disabilities Act (ADA). Have an HR professional review every posting before it goes live, and consider legal review for roles with specific credential or physical requirements. ChatGPT can help you rewrite a phrase that legal flags, but it cannot make the legal judgment itself.

Reviewing and updating HR policies

Pasting a policy document into ChatGPT and asking it to identify confusing language, check for internal consistency, or compare it against a plain-language standard is a legitimate use. ChatGPT is reasonably good at spotting vague pronouns, contradictory provisions, and sentences that require multiple readings. Give it a specific task rather than a general "improve this."

Do not ask ChatGPT to verify whether a policy complies with current law. It will give you a plausible answer, but it cannot reliably know whether your state passed a new employment law last month. Use ChatGPT to improve clarity and structure; verify legal compliance with your employment counsel or against official sources like the Department of Labor's guidance at dol.gov.

Preparing onboarding and internal communications

Onboarding communications follow predictable patterns: welcome message, first-week checklist, benefits summary, IT setup instructions. ChatGPT can draft all of these from your notes. The output needs review against your current policies and benefits before it goes out, since template language can drift from what is actually true for your plan.

For internal announcements, role changes, or departures, ChatGPT can help with tone and structure. Give it the key facts you want to communicate, the audience, and any constraints (what not to say, what to be careful about), and ask for a draft. Review carefully: an announcement about a sensitive change should not have a tone that ChatGPT thought was appropriate but your team knows is wrong.

Interview preparation and training materials

ChatGPT can draft structured interview question sets from a job description and a set of competencies. Give it the role, the competencies you are hiring for, and any constraints (behavioral questions only, no hypotheticals), and it will produce a starting set. Review for relevance, legality, and fairness before you use them.

For training materials, ChatGPT works well as a drafting tool for scenario-based content: "draft five training scenarios for new managers about delivering difficult feedback." The scenarios will be generic without your context; add specifics from your workplace and have a subject-matter expert review before they go into a training session.

What ChatGPT should not do in HR work

Several categories of HR work should stay with people. Do not use ChatGPT to evaluate individual employees, score candidates, or produce any output that directly affects an employment decision without a human making that decision consciously. Do not use it to produce final discipline documentation or termination letters without legal review. Do not ask it questions that require knowledge of a specific employee's situation.

ChatGPT is not appropriate for tasks that require accessing or processing employee personal information. Even if you use an enterprise plan with data protection agreements, the practice of pasting employee records into a chat interface creates risk. Keep individual employee information where it belongs: in your HR system.

Protecting employee information

The most important HR rule for ChatGPT use is simple: do not put employee data into it. That means no names with performance details, no salary information, no medical information, no details from a specific employee's record, and no disciplinary content that identifies an individual. Draft templates without names; fill in specifics in your HR system.

If you use ChatGPT through a paid enterprise plan, review OpenAI's current data handling terms at openai.com. Plans differ in how prompts are stored and whether they are used for training. Check what applies to your account type before using it for HR work. For ChatGPT alternatives and how ChatGPT compares to other tools, those pages cover adjacent decisions. For Claude vs. ChatGPT and the related workflow automation tools, see those comparisons.

How to do this with fenbs

Set up an HR project in fenbs to keep your ChatGPT workflow rules and prompts in one shared place.

1. Create a free account and add a project for your HR team's AI work.
2. Write the ground rules in Decisions and rules. Cover what ChatGPT can draft, what must go to legal before use, and what information must never go into ChatGPT.
3. Add your approved starting prompts to Skills and prompts. Include one prompt per document type: job description, policy review, onboarding checklist.
4. Connect your AI assistant over MCP from Integrations. It reads the Decisions and rules at the start of each session.
5. When a prompt or rule changes after review, update it in fenbs. The next session starts from the current version.