Texas HR Consulting for High-Stakes People Decisions.
Return to HR FAQ Library

Can managers use ChatGPT to write employee discipline?

With rules, yes. Without them it produces confident, generic, unverifiable documentation — which is worse than none.

Last updated: August 02, 2026

Direct Answer

Managers should not paste employee names, health information, complaint details, or other identifying information into public AI tools, and no AI-drafted discipline should be issued without the manager verifying every stated fact. Used with those two rules, AI drafting assistance is defensible: it can structure a write-up while the manager supplies the facts. Used without them, it creates confidentiality breaches and documentation that falls apart under scrutiny.

Controlling authority: Title VII and the ADA, which govern by disparate impact regardless of intent, and TRAIGA (Tex. H.B. 149), which is intent-based and enforceable only by the Texas Attorney General.

The Two Failure Modes

The first is data leakage. A manager who pastes a complaint summary with names into a public chatbot has disclosed confidential personnel information to a third party, potentially including health details that carry their own legal protections. Public AI tools are outside your control: what goes in cannot be recalled, and some tools use inputs for training.

The second is fabricated precision. Generative tools produce confident, specific prose, and a manager in a hurry will sign a write-up describing three prior conversations that never occurred or policy language your handbook does not contain. Documentation is evidence. A write-up with invented facts is worse than no write-up, because it proves the process cannot be trusted.

The sections an HR AI use policy has to contain Faulkner HR Solutions. Original framework, 2026.
SectionWhat it must sayThe failure it prevents
Permitted and prohibited usesNamed tools, named tasks, and the tasks that are off limits“Use good judgement” is not a policy.
Data classificationWhat may never be entered: names, medical information, complaint details, pay data, anything from a personnel fileThe single largest exposure, and the easiest to prevent.
Human accountabilityA named person owns every output; AI is never the author of a decisionRemoves the “the tool said so” defence before anyone tries it.
Verification dutyEvery factual or legal statement is checked before useFabricated citations and invented policy language.
Record statusAI drafts are discoverable, and prompts may be tooManagers assume the draft is private. It is not.
Vendor and tool approvalWho approves a new tool, and on what criteriaShadow adoption of unvetted tools.
Review cadenceWhen the policy and the tool list are revisitedA policy naming tools that no longer exist.

The Rules That Make It Workable

Set a bright line on data: no employee names, identifiers, health information, or complaint details in public AI tools, ever. If the organization wants AI drafting help, provide an approved tool with appropriate data terms, or require managers to draft with placeholders and add specifics afterward.

Require the manager to own every word. The AI can propose structure: incident, expectation, prior notice, consequence. The manager must verify each factual claim against records before anything is issued or signed, and HR review before issuance catches the rest. The signature line belongs to a human for a reason.

AI Drafting Risks to Watch

These problems are invisible until a write-up is challenged or a breach surfaces. Watch for these.

  • Employee names and complaint details pasted into public chatbots
  • Write-ups citing conversations or policies that do not exist
  • Discipline language inconsistent with how similar cases were documented
  • Managers issuing AI-drafted documents without HR review
  • No policy telling managers any of this

Set the drafting rules before the next write-up

Ask your managers, without accusation, whether they use AI for people documentation. The honest answer is usually yes, which means the policy conversation is overdue rather than hypothetical.

Spot-check recent write-ups against source records: do the cited dates, conversations, and policy sections check out? Errors found now are corrections; errors found in litigation are impeachment.

Free tool

Documentation Defensibility Scorecard

Rates a real file against the standard a reviewer will apply to it.

When AI-drafted records are already in files

Get an AI use policy in place before the first incident rather than after, and pair it with a one-hour manager briefing on what changes and what does not.

If a write-up with AI-invented facts has already been issued and challenged, get guidance before responding, because the correction path matters.

Get a Straight Answer for Your Situation

General rules only go so far. If this question is live in your organization right now, talk it through with a senior HR consultant before you act. One conversation now costs less than one claim later.

Contact Us

Written and reviewed by Dr. Thomas W. Faulkner, DBA, MBA, MSML, SPHR, LSSBB, principal consultant at Faulkner HR Solutions, a Texas HR consulting firm based in San Antonio serving small businesses, nonprofits, municipalities, and public sector employers.

This page provides general HR information for employers and is not legal advice. For legal interpretation or representation, consult qualified employment counsel.