Cybernetic NetworksTHE TECH BRIEF.
Tech Tips

Don’t get left behind: use AI safely in your business.

Half of small business workers already use AI. Here's how to get the benefits without handing your customer data to strangers, and how scammers are using the same tools against you.

Person typing a question into an AI assistant on a laptop

This morning, one of your competitors got an hour’s work done before their first coffee went cold. A customer complaint answered politely. A 20-page contract boiled down to the five points that matter. A stubborn Excel problem solved in two minutes. They didn’t hire anyone new. They used an AI assistant.

This isn’t the future anymore. Three out of four small business owners in a national survey this year said they already use AI in some form. If you haven’t started yet, you’re not alone and you’re not too late. But the gap between businesses that use it and businesses that don’t is growing every month.

There is a catch, and it’s the reason many owners hold back: whatever you type into an AI tool leaves your computer. Used carelessly, it can leak customer details or company information. And criminals are using the very same technology to write perfect scam emails and even fake the voices of people you trust.

This article covers both sides. How to put AI to work in your business this week, what it really costs and why it pays for itself, exactly where your information goes when you type into it, and the simple habits that keep your customers’ data and your money safe.

By the numbers

AI is already in your competitors’ offices

  • 76% of small business owners surveyed by Goldman Sachs 10,000 Small Businesses in early 2026 said they use AI in some form. Only 14% had fully built it into how they run the business.
  • About half of small business workers now use AI for their work, according to a U.S. Chamber of Commerce Foundation survey from May 2026.
  • 77% of employees who use AI paste company information into it, and most of that pasting happens in personal accounts the business doesn’t control (LayerX, 2025).
  • 22,000+ complaints about scams involving AI were reported to the FBI in 2025, with losses close to $893 million.

Why sitting this out is the riskier choice

Many business owners I talk with are holding back on AI for understandable reasons: they don’t trust it, they don’t have time to learn it, or they’ve heard it can leak information. Those concerns are fair. But waiting carries its own cost.

Your competitors are answering customer emails in half the time. They’re writing job listings, website text and social posts in minutes instead of evenings. Their staff are getting instant answers to “how do I do this in Excel?” instead of waiting for someone to help. In the Goldman Sachs survey, 93% of the owners using AI said it had a positive effect on their business. And the biggest reason most hadn’t gone further wasn’t fear. It was not knowing where to start.

There’s a second, quieter reason not to wait: your employees are probably already using AI, whether you’ve approved it or not. Study after study finds that a large share of workers use AI tools on their own, often with free personal accounts. If you haven’t chosen a tool and set some rules, you have no say in what company information they type into it. Doing nothing doesn’t keep AI out of your business. It just means it’s happening without you.

What AI actually does in a normal business day

Forget the science-fiction headlines. Today’s AI assistants, such as Microsoft Copilot, ChatGPT, Claude and Google Gemini, are best thought of as a fast, well-read assistant who never gets tired of a question. Here’s what that looks like in practice:

At the front desk

Turning a rushed reply into a polite, professional one. Calming down an angry customer email without sounding defensive. Translating a message for a customer who speaks Spanish. Creating a list of answers to the ten questions customers ask every week.

For the owner

Summarizing a 30-page contract or lease into the points that matter, and listing questions to ask your attorney. Drafting a difficult letter to a vendor. Thinking through a decision: “What are the pros and cons of raising our prices 5% in January?”

In the back office

Explaining an Excel formula, or writing one from a plain-English description: “add up column D only where column B says Paid.” Turning messy meeting notes into a clean list of who’s doing what. Writing step-by-step instructions so a new hire can learn a task.

For marketing

Writing a month of social media posts from a few notes. Improving the wording on your website. Coming up with ideas for a holiday promotion. Writing a job listing that actually attracts good applicants.

In each case, the AI does the first draft and you make the final call. That’s the right way to think about it: a helper, not a decision-maker.

“The question isn’t whether your team will use AI. It’s whether they’ll use it safely.”

Where your words go when you hit “Enter”

Here’s the part most people never think about. When you type a question into an AI tool, your words don’t stay on your computer. They travel over the internet to the company that runs the AI, where they’re processed and often saved in your chat history.

What happens next depends almost entirely on which kind of account you’re using:

  • Free and personal accounts are built for individuals. With ChatGPT’s free and personal paid plans, for example, OpenAI says your conversations can be used to train its AI unless you switch that setting off. The protections, controls and contracts are aimed at consumers, not businesses.
  • Business accounts are built for companies. The major providers state that business customers’ information is not used to train their AI, and they give the business owner control over who has access.

Why does “training” matter? Training is how an AI learns. Information used for training becomes part of what the system has learned from. You don’t want your customer list, your pricing or your employee’s medical note to be part of that.

This isn’t a theoretical problem. In 2023, engineers at Samsung pasted confidential source code and internal meeting notes into ChatGPT to get help with their work. When the company found out, it temporarily banned staff from using AI tools. Nobody meant any harm. They were just trying to get their jobs done faster, which is exactly how these leaks happen.

Security researchers have a name for this: “shadow AI,” meaning AI tools used for work without the company knowing or approving. Netskope, a security company that monitors business internet traffic, found that nearly half of people using AI at work were doing it through personal accounts their employer couldn’t see. Its 2026 report counted an average of more than 200 incidents a month per organization in which sensitive information was sent to AI apps.

Good news if you have Microsoft 365

You may already have a safe AI tool and not know it

If your business uses Microsoft 365 for email, your team can use Microsoft Copilot Chat by signing in with their work email account, at no extra cost. When you’re signed in with a work account, Microsoft says it applies “enterprise data protection”: your prompts and the files you add aren’t used to train its AI models. Look for the shield symbol or the word “Protected” near the chat box. If you don’t see it, you’re probably signed in with a personal account.

Google offers similar protection for Gemini in Google Workspace business accounts, and the business plans of ChatGPT (Business and Enterprise) and Claude (Team and Enterprise) state that they don’t train on business data by default.

What it costs, and why it pays for itself

Most business owners I talk with ask the same first question: “What’s this going to cost me?” It’s the right question. The good news is that the answer is a lot less than most people expect.

The two best-known AI assistants both sell business plans at nearly the same price. ChatGPT Business costs $20 per person per month on a yearly plan, or $25 if you pay month to month, with a minimum of two people. Claude Team from Anthropic is priced the same way: $20 per person per month yearly, or $25 monthly, for teams of 2 to 150 people. If you already use Microsoft 365, the full Microsoft 365 Copilot Business add-on, which works inside Outlook, Word, Excel and Teams, lists at $21 per person per month on a yearly commitment. And remember, the basic Copilot Chat is already included with your work account at no extra cost.

So for most businesses, the real figure is about $20 to $25 per person per month. That’s less than a lunch meeting.

The simple math

If it saves one hour a month, it has paid for itself

Say an employee costs you about $25 an hour once you add payroll taxes. An AI seat costs about $20 to $25 a month. If AI saves that person just one hour in an entire month, the seat has paid for itself.

In practice, people who use AI every day save far more than that: a faster reply here, a summarized document there, an Excel problem solved in two minutes instead of twenty. Here’s what it looks like for a team of five:

  • Cost: 5 seats × $25 = $125 a month
  • Time saved: if each person saves just 1 hour a week, that’s about 20 hours a month
  • Value of that time: 20 hours × $25 = $500 a month

That’s roughly four dollars back for every dollar spent, and that’s using conservative numbers. Those hours don’t disappear: they go toward customers, sales and the work that actually grows your business.

My pick

Which one would I choose? Claude.

People ask me this all the time, so here’s my honest opinion. ChatGPT is a good tool and millions of businesses use it well. But for day-to-day business work, I personally prefer Claude, made by Anthropic.

  • It writes like a person. Emails and letters come out sounding natural and professional, with less editing needed before you send them.
  • It handles long documents well. Contracts, policies and long reports are easy to load in and ask questions about.
  • It’s careful. In my experience it’s more likely to say when it isn’t sure, instead of confidently making something up.
  • Team workspace. With Claude Team, your staff can share projects and company information so everyone works from the same instructions.

Whichever you choose, the rules in this article are the same. And if you already pay for Microsoft 365, start with the free Copilot Chat on your work account. You’ll quickly see whether it’s worth adding more.

Why not just pay for employees’ personal accounts?

Some owners think, “I’ll just reimburse my staff for a personal $20 subscription.” It costs about the same, so why does it matter? Because with a business plan, the company owns the work, not the employee. Here’s what the business version gives you that personal accounts don’t:

  • A shared company workspace. Your team can share projects, saved instructions and useful conversations with each other. Build a “customer reply helper” once, and everyone uses it the same way.
  • Your work stays with your business. When an employee leaves, you remove their seat and everything they built stays in the company workspace. With a personal account, months of useful work walks out the door with them, along with any company information they typed in.
  • Your data isn’t used to train the AI. Both ChatGPT Business and Claude Team state that business conversations aren’t used to train their models by default.
  • You’re in control. One bill, one place to add or remove people, and sign-in with your company email accounts.
  • Consistent quality. Everyone works from the same tools and the same instructions, so your customers get the same quality of response whoever writes it.

There’s a less obvious benefit too. When you give your team a proper tool and permission to use it, they stop sneaking around with free accounts. You get the productivity boost and the control, and that’s the whole point.

The business case in one sentence
About $20 to $25 a month per person, and it pays for itself if it saves each person a single hour a month. Many people save that in their first week.

The 7 rules for using AI safely

These are the rules I recommend to every business owner. None of them take more than a few minutes to put in place.

1. Use a business version, not a free personal one

This is the single most important step. A business account gives you a written promise about how your data is handled, and it gives you control. For most Microsoft 365 businesses, the easiest starting point is Copilot Chat with your work account, as described above.

2. Sign in with your work account

When everyone uses company accounts, the company stays in charge. When an employee leaves, you turn off one account and their access to the AI, and to everything they discussed with it, ends with it. With personal accounts, that history walks out the door with them.

3. Never paste sensitive information

Even with a business account, some things should never go into an AI tool. See the traffic-light guide below. When in doubt, leave it out.

4. Remove names and numbers before you paste

AI doesn’t need real details to do a good job. Instead of “Write a reminder to John Smith at Smith Plumbing, who owes us $4,250 on invoice 10432,” write “Write a friendly reminder to a customer whose invoice is 30 days overdue.” You’ll get the same quality letter. Then add the real details yourself before you send it.

5. If you use a free tool, turn off training

If you do use a personal account for something general, change the privacy setting. In ChatGPT, it’s under Settings → Data controls → “Improve the model for everyone,” which you should switch off. In Claude, it’s under Settings → Privacy → “Help improve Claude,” also switched off. Other AI apps have a similar setting in their privacy or data section. Your chats will still be saved in your history, but OpenAI says new conversations won’t be used for training.

6. Always check the answer

AI can sound completely confident and still be wrong. It can invent a statistic, misread a contract or get a date wrong. Read everything before you send it. Double-check numbers, names, dates, prices, and anything legal, medical or financial. You’re responsible for what goes out under your name, not the AI.

7. Put your rules in writing

A one-page AI policy tells your team which tools are approved, what never goes in, and who to ask. Most problems happen because nobody said anything, not because someone broke a rule. There’s a simple template at the end of this article.

Traffic-light guide

What can go into an AI tool?

🟢 GREEN, fine to use: general questions, public information, your own rough notes with names removed, website and marketing text, Excel and software questions, ideas and brainstorming.

🟡 YELLOW, business account only, and remove names first: customer emails, internal procedures, meeting notes, pricing ideas, draft contracts or proposals.

🔴 RED, never: Social Security numbers, credit card or bank account numbers, passwords and login codes, medical or health information, employee personnel files, anything a customer gave you in confidence, and anything covered by a confidentiality agreement.

The other side: scammers are using AI too

For years, the easiest way to spot a scam email was bad spelling and strange grammar. That clue is gone. AI now writes scam emails in perfect English, in your bank’s tone of voice, sometimes using details pulled from your website or social media.

Voices can now be copied too. With just a short clip of someone speaking, taken from a video on Facebook, a voicemail greeting or a local news interview, criminals can create a voice that sounds like you, your business partner or your grandchild. The FBI received more than 22,000 complaints involving AI in 2025, with reported losses of nearly $893 million, and it has repeatedly warned the public about cloned-voice calls and messages.

How convincing can it get? In early 2024, a finance employee in the Hong Kong office of Arup, the global engineering firm behind the Sydney Opera House, joined a video call with the company’s chief financial officer and several colleagues. They discussed a confidential deal that needed money moved quickly. He had been suspicious of the email that started it all, but seeing and hearing his own bosses put his mind at ease. He made 15 transfers totaling about $25.6 million.

None of the people on that call were real. Hong Kong police later said every face and voice had been created with AI, built from videos of the real executives found online. The fraud was only discovered when the employee checked with head office afterwards. He did several sensible things: he doubted the first email and wanted to see the people involved. It still worked, because seeing a familiar face felt like proof. These scams don’t need perfect technology. They need a moment of trust and a sense of urgency.

The one rule that beats every deepfake
Never move money, change payment details or share a password because of one message, call or video, no matter who it seems to be from. Hang up and call back on a number you already have.

How to protect your business and your family

  • Call back on a known number. If “the boss,” a vendor or a relative asks for money or information, end the conversation and call them on a number you already have, not one they gave you.
  • Use a code word. Agree on a simple word with your family and your key staff. If someone calls in a panic asking for money, ask for the word. An AI copy of a voice won’t know it.
  • Require two people for payments. Any wire transfer, or any change to a vendor’s bank details, should need a second person to approve it. Scammers rely on one person acting alone under pressure.
  • Treat urgency and secrecy as red flags. “Do this now” and “don’t tell anyone” are the two phrases almost every scam uses.
  • Never change payment details by email alone. If a vendor says their bank account has changed, confirm it by phone with someone you know there.

A one-page AI policy you can copy

Print this, fill in the blanks, and go over it at your next staff meeting. It takes ten minutes and prevents most of the problems in this article.

Template

[Company name] AI Use Policy

1. Approved tools: We use ________ (for example, Microsoft Copilot Chat), signed in with your company account only. Don’t use personal AI accounts for company work.

2. Never enter: customer or employee personal information, Social Security, card or bank numbers, passwords, health information, or confidential documents.

3. Remove names: take out names, account numbers and addresses before asking AI for help with a customer matter.

4. Check everything: you are responsible for anything AI helps you write. Read it, and check facts, numbers and dates before sending.

5. Money and passwords: never act on a request to pay, change bank details or share a login based on one email, call, text or video. Always confirm by calling a known number.

6. Questions: ask ________ before using a new AI tool or app.

Five safe ways to start this week

The best way to get comfortable with AI is to use it on small, low-risk jobs. Try these, word for word if you like:

  1. Fix a tricky email. “Rewrite this email so it’s polite and professional but still firm: [your draft with names removed].”
  2. Write a job listing. “Write a job listing for a part-time receptionist at a [type of business]. Duties: answering phones, scheduling, greeting customers. Friendly tone, under 200 words.”
  3. Solve an Excel problem. “In Excel, how do I add up the amounts in column D only for rows where column B says ‘Paid’?”
  4. Plan a promotion. “Give me 5 ideas for a January promotion for a [type of business] that wants more repeat customers.”
  5. Summarize something long. Paste in a public article or a long policy and ask: “Summarize this in 5 bullet points a busy owner can read in 30 seconds.”

After a week of this, you’ll have a good feel for what AI does well and where it needs checking. Then you can look at bigger jobs, such as answering common customer questions or drafting proposals.

The bottom line

AI isn’t a fad, and it isn’t only for big companies. It’s quickly becoming as ordinary as email, and the businesses that learn to use it well will save hours every week. The ones that ignore it will end up with employees using it anyway, without any rules.

The safe path is simple: choose a business version, sign in with work accounts, keep sensitive information out, check what it gives you, and teach your team to verify any request for money with a phone call. Do those five things and you get the benefits of AI without handing your data, or your bank account, to strangers.

Want help choosing and setting up AI the safe way? We help local businesses pick the right AI tools, set them up with proper work accounts and security settings, and write a simple AI policy for the team. Call or text us at 407-554-5534.

Sources

  • CNN and Fortune, reporting on the Arup deepfake video call fraud (May 2024), citing Hong Kong police.
  • Goldman Sachs 10,000 Small Businesses Voices, survey of 1,256 small business owners (January–February 2026).
  • U.S. Chamber of Commerce Foundation, Main Street AI Monitor, conducted by Ipsos (May 2026).
  • LayerX, Enterprise AI and SaaS Data Security Report (2025).
  • Netskope, Cloud and Threat Report (2026).
  • FBI Internet Crime Complaint Center (IC3), 2025 Internet Crime Report.
  • OpenAI Help Center, “ChatGPT Business overview” (plan pricing).
  • Anthropic, Claude Team plan pricing (claude.com/pricing).
  • Microsoft, Microsoft 365 Copilot Business pricing.
  • Microsoft Support, “Data protection when using Microsoft Copilot Chat for work or school.”
  • OpenAI Help Center, “Data controls in ChatGPT.”
Free IT & Security Review

How protected is your business, really?

We'll look at your computers, email, backups and network, and tell you in plain language what's working and what isn't. No cost, no pressure.