AI Chatbots at Work: What Your Staff Are Typing, and Where It Goes
Updated: 2 days ago
Your employees are almost certainly using AI chatbots at work, and on consumer accounts what they type can be kept and used to train the models of the company behind the tool. That puts client details, pricing and internal documents at risk. You do not need to ban AI to fix it. You need a written policy, approved tools on business plans, and a little training.

What the research found
In October 2025, researchers at Stanford reviewed the privacy policies of six major U.S. AI developers: Amazon, Anthropic, Google, Meta, Microsoft and OpenAI. All six use the chats people have with them to train their models by default. Some let users opt out; some do not. (Stanford HAI, October 15, 2025)
The lead researcher, Jennifer King, advises people to think twice about what they share with a chatbot and to opt out of training wherever they can.
Why this matters for a business
When an employee pastes something into a personal chatbot account, the business has no say in what happens to it. Three common ways it goes wrong:
Confidential work leaves the building. A draft contract, a client list or a pricing sheet pasted in to "clean it up" now sits with an outside company under consumer terms.
Regulated data goes where it should not. Patient details, client financial records or employee information typed into a consumer tool can create a compliance problem the moment it is sent.
Nobody can see it happening. Personal accounts and personal phones are invisible to your normal IT controls.
Business plans are different
Most AI vendors offer business and enterprise plans that do not train on your data by default and give an administrator control over who uses them. Those protections come with the business plan, not the free account, and the terms change often. Check the current terms before you approve a tool.
A 30 day plan
Week 1. Find out what is in use
Ask your team which AI tools they use and what for. Keep it friendly; the goal is a picture, not a reprimand. Then write a short AI acceptable use policy. Our guide explains what goes in one.
Week 2. Pick approved tools
Choose one or two tools on business plans and make them the approved path. Block or discourage the rest on company devices and networks.
Week 3. Train your people
Give real examples of what never goes into a chatbot: client names, account numbers, health details, passwords, unreleased pricing. Explain why. People follow rules they understand.
Week 4. Review and repeat
Check how the approved tools are being used, update the policy, and put a review on the calendar. AI tools and their terms change every few months.
Should we just ban AI chatbots?
A blanket ban is hard to enforce and pushes use onto personal phones, where you have even less visibility. A better rule is simple: approved tools for work, nothing else, and nothing sensitive in any tool without approval.
Common questions
Do AI chatbots keep what we type?
On most consumer accounts, yes, and it may be used to train the model unless you opt out. Business plans usually work differently, so check the terms for the plan you are on.
What should employees never put into a chatbot?
Client and patient details, financial account information, passwords, employee records, and anything covered by a confidentiality agreement.
Can AllTech help us roll out AI safely?
Yes. We help clients choose approved tools, write the policy, set up the controls and train staff. See our AI, Agents and Automation services.
Get a plan for AI at your business
Talk to our team at (205) 290-0215, or see your price on our pricing calculator.

Comments