What are the risks of using AI in a small business?
The main risks are wrong answers, private data sent to an outside company, a customer misled, a promise made in your name, copied content, and a tool you come to depend on that later changes. Each can be limited: a person checking what customers see, a rule about what goes in, written boundaries, and a way to switch it off. A limit lowers a risk but does not remove it.
The risks and what limits each
| Risk | What limits it | Where it is covered |
|---|---|---|
| Wrong answers | Approved information only, permission to say it does not know, and a person checking anything a customer may hold you to. This lowers the risk but does not remove it. | The related answer on stopping an AI assistant from making things up |
| Private data sent out | Leave sensitive data out, and read the data terms of the plan you use. | The related answer on putting customer information into ChatGPT |
| A customer misled or put off | A person reads messages before they go, and you say plainly where you use AI. | The related answers on approving what AI sends and on telling customers |
| A promise made in your name | A written never-say list, and a handoff to a person. | The related answer on what an AI assistant should never say |
| Copied content | Avoid generated images that look like a real person, a known character or another business's logo, read the tool's terms on business use, and check text before you publish it. Ask a lawyer if you are unsure. | The related answer on using AI-generated images on your website (images only) |
| A tool you depend on changes | Accounts in your name, a way to switch it off yourself, and the manual way written down. | The related answer on what to ask before hiring an AI builder |
General advice before you rely on a tool
This section is general advice, not drawn from the sources cited here. Before a tool touches real work, have a second person look at what it produces, read what its plan does with what you type, and decide who in the business may use it. Messages and campaigns deserve a second look in particular, because AI-written text can read as generic or get a detail about a customer wrong. For a question about copyright, privacy law or a contract, ask a lawyer; this is general information, not legal advice.
What NIST adds
NIST says it developed a framework, released on January 26, 2023, to better manage risks to individuals, organizations and society associated with AI, and that it is "intended for voluntary use". So NIST presents it as voluntary guidance, not a requirement. Its Generative Artificial Intelligence Profile, released on July 26, 2024, can help organizations identify unique risks posed by generative AI, and NIST says version 1.0 of the framework is being revised. The table above is this answer's own, not NIST's.
What none of this removes
A limit lowers a risk; it does not remove it. Keep a person on anything a customer may hold you to, and check a sample of the tool's work often. The related answer on where to start with AI keeps the first task small.
Sources
Related answers
- How do you stop an AI assistant from making things up?
- Is it safe to put customer information into ChatGPT or another AI tool?
- Should a person approve what an AI system sends to customers?
- What should an AI assistant never say about your business?
- Do you have to tell customers when you use AI to answer or write to them?
- What should you ask before hiring someone to build AI for your business?
- Where should a small business start with AI?
- Can you use AI-generated images on your business website?
AI systems and automation
A lot of an owner's week goes to the same tasks: answering the same questions, chasing the same follow-ups, retyping the same details. We build the systems that do that work, so your people can spend their time on customers.