AI Compliance Kit: Is ChatGPT, Claude or Grok Safe for Client Work?

1. What you get

Your staff are already using AI. The real question for a law firm or professional office is whether the AI vendor's settings, and what you tell clients and employees, make it safe to use for client work. The kit answers that for OpenAI (ChatGPT and the API), Anthropic (Claude and the API) and xAI (Grok and the API). You get:

A plan-by-plan verdict. The API, and each chat plan (for example ChatGPT Business vs. Enterprise, or Claude Team vs. Enterprise), rated for client work.
The exact settings to change. Read from your vendor account, not guessed.
What recent court decisions on AI and privilege mean for you.
Ready-to-adapt language for your engagement letter, employee handbook and a client advisory.

The report is saved as a document in the kit's reports folder.

Already use docMonster? The same assessment is built in: open docMonster's Settings, go to the AI Keys tab and click Compliance…. The kit is for anyone who wants to run it without installing docMonster.

2. What you need

An AI coding assistant:Claude Code, OpenAI Codex or Grok Build, installed and signed in. Use a business or API account, not a personal one: what you share with it is sent to that AI company.
Python 3.8 or later, free from python.org. On Windows, tick Add python.exe to PATH during setup. The assistant uses it to read your vendor settings and your documents. Without it you can still run the assessment, with less detail.
Optional, recommended: a read-only admin key for the AI vendor you're assessing (the assistant shows you how to create one), plus your employee handbook and engagement letter template.

3. Run the assessment

Download and unzip the kit somewhere easy to find, such as Documents\AI Compliance Kit.
Open a terminal in that folder. Windows: right-click the folder in File Explorer and choose Open in Terminal. Mac: open Terminal, type cd , drag the folder onto the window and press Return.
Start your assistant: claude for Claude Code, codex for OpenAI Codex, or grok for Grok Build (answer Yes when it asks whether to trust the folder).
Type: Run the compliance assessment

From there the assistant leads. It asks what kind of firm you are, which vendor and chat plans you use, and your state. It shows you where to create a read-only admin key and has you paste it into a settings file yourself, then reads your account settings. It asks you to drop your handbook and engagement letter into the kit's documents folder, and writes the report. If the assistant asks permission to connect to the internet, approve it: that's how it reads your vendor settings.

4. Keeping it private

Your admin key goes in a file named .env in the kit folder. You paste it there yourself in Notepad or TextEdit. Never paste a key into the chat.
Only the included script reads the key, and it never prints it. The key only needs read-only access.
When you're done, delete the admin key in the vendor's console and delete the .env file.
Don't put documents that contain client information in the documents folder. Your handbook and a blank engagement letter template are all it needs.

5. Questions

Does it cost anything?

The kit is free. The assistant runs on your own Claude, ChatGPT or Grok subscription or API credits.

Is the report legal advice?

No. It's an automated review of your vendor settings and documents. Have it reviewed by your compliance counsel before you rely on it.

How current is the vendor information?

The kit's vendor reference is current as of October 2026, and the assistant reads your live account settings each time you run it. Download the kit again for updates: the link always points to the latest version.

AI Compliance Kit