ChatGPT for Small Business: Useful Tool, Real Risks

Three and a half months ago, a company called OpenAI released a free chatbot called ChatGPT. Within weeks, it was all anyone could talk about. People were using it to write poems, plan vacations, debug code, and draft cover letters. Today OpenAI announced GPT-4, a more capable model behind the tool, and the buzz is only getting louder.

We are getting a lot of questions from business owners about it. Is it useful? Is it safe? Should employees be using it at work? The honest answer is yes, it can be genuinely useful, and yes, there is one big risk you need to get ahead of right now.

What ChatGPT actually is

ChatGPT is what is called a large language model. It was trained on an enormous amount of text, and it learned to predict what words should come next. The result is a tool that can hold a conversation, answer questions, and produce writing that sounds remarkably natural.

What it is not is a search engine or an expert. It does not "know" things the way a person does, and the free version has limited knowledge of recent events. It can be confidently wrong, sometimes inventing facts, statistics, or even sources that do not exist. Keep that in mind and it becomes a very handy assistant. Forget it and it can embarrass you.

Treat ChatGPT like a fast, eager new hire: great for a first draft, never the final word.

Where it genuinely helps a small business

Here are uses we think make real sense for a typical small office today:

  • First drafts of routine writing. Customer emails, follow-up letters, job postings, meeting agendas, and social media posts. Give it the key points and tone you want, then edit the result.
  • Rewriting and polishing. Paste in a rough paragraph you wrote and ask it to make it clearer, friendlier, or shorter. This alone saves many people a lot of time.
  • Brainstorming. Ideas for a newsletter, names for a new service, questions to ask in an interview, or the outline of a hurricane preparedness checklist for your staff.
  • Explaining things. Ask it to explain an unfamiliar term, a spreadsheet function, or how a type of business insurance generally works. Double-check anything important.
  • Spreadsheet help. Describe what you want a formula to do in plain English, and it will usually suggest one. Test it on sample data before trusting it.
  • Summarizing public information. Long articles or publicly available documents can be condensed into a few bullet points.

The pattern is the same in every case: it gets you from a blank page to something workable, and a human finishes the job.

The big risk: what you paste in

This is the part we most want business owners to hear. When someone types or pastes information into the public version of ChatGPT, that information leaves your business and goes to OpenAI's servers. OpenAI's own terms say conversations may be reviewed and may be used to improve its models, and it tells users not to share sensitive information.

That means employees should not be pasting in:

  • Customer names, contact details, or account information
  • Patient information, if you are in healthcare, or anything else covered by privacy regulations
  • Financial records, payroll, or bank details
  • Contracts, legal documents, or anything under a confidentiality agreement
  • Passwords, internal network details, or proprietary code
  • Internal emails or documents that you would not post publicly

The trouble is that it is so easy. An office manager wants help summarizing a messy customer complaint and pastes the whole email thread in, names and all. A bookkeeper asks for help analyzing a spreadsheet and pastes in the actual numbers. Nobody means any harm, and most people simply have not thought about where that text goes.

Other things to watch

  • Accuracy. Never send customers facts, figures, or legal or medical guidance straight from ChatGPT without checking them.
  • Your voice. AI-generated text can sound generic. Customers who know you will notice if your emails suddenly sound like a brochure.
  • Scammers use it too. The same tool that helps you write a clean email helps criminals write convincing phishing messages without the usual spelling mistakes, as we covered in our post on AI-powered phishing.
  • Fake apps. Its popularity has spawned look-alike apps and browser extensions, some of which are simply there to harvest data. Stick to OpenAI's official site.

Put a simple policy in place now

You do not need a thick policy manual. You need a few clear rules that everyone understands, written down before habits form. Some employees are almost certainly already using ChatGPT, whether or not anyone has said it is allowed.

  1. Decide whether and how staff may use ChatGPT and similar tools for work, and tell them.
  2. Make the "never paste" list above part of the rule, in plain language.
  3. Require a human to review anything AI-assisted before it goes to a customer.
  4. Use it only through the official website, and use work accounts with strong, unique passwords.
  5. Keep an eye on the business versions and privacy controls that are likely to arrive as these tools mature.
  6. Revisit your rules in a few months, because this technology is changing quickly.

The bottom line

ChatGPT is not magic, and it is not a threat to be feared. It is a useful new tool that can save your team real time on everyday writing and thinking tasks, as long as people use it with good judgment and keep confidential information out of it. The businesses that do well with it will be the ones that set sensible ground rules early.

If you would like help writing an acceptable-use policy for AI tools, or reviewing how data moves in and out of your business, our cybersecurity team is glad to help. Reach out at info@pinellasit.com or through our contact form.

Need help with this?

Talk to a local IT team that has supported businesses across Pinellas & Pasco County since 2010.

Get a Free Consultation Free Security Assessment