Published On: November 2nd, 20265.3 min read
Laptop beside stacked folders and a padlock shield icon, representing protecting business data when using AI tools

Here’s a likely scene at your business this week: someone pastes a client email thread into a free AI chatbot to draft a reply. Someone else uploads a spreadsheet to get help with formulas. A manager installs a browser extension that promises to summarize every web page.

None of them meant any harm. But each of those moments can send business data to a service you never vetted, under terms you never read. For most small businesses, the biggest AI data risk isn’t a sophisticated attack on an AI model. It’s everyday use of AI tools with no rules.

Where AI data risk really comes from for small businesses

Large companies worry about attacks on the AI systems they build. Most small businesses don’t build AI. They use it. That shifts the risk to a few practical places:

  • Personal or free AI accounts. Staff sign up with personal email, and business data ends up in an account you don’t control and can’t shut off when they leave.
  • Unclear data use. Some consumer AI services may use what you type to improve their products unless you change a setting. Business and enterprise versions often have different terms. You need to know which one your team is using.
  • Browser extensions and plug-ins. Many AI add-ons request broad access to read every page you visit, including your email and banking.
  • Oversharing inside your own systems. AI assistants built into business software can surface files people technically have access to but were never meant to see.
  • Account takeover. If an attacker gets into an employee’s account, any AI chat history stored there may be readable too.

Step 1: Find out what’s already in use

You can’t set rules for tools you don’t know about. Spend a week getting a picture:

  • Ask your team, without judgment, which AI tools they use and for what.
  • Have your IT provider review which third-party apps are connected to your Microsoft 365 accounts.
  • Check browser extensions on company computers.

Expect surprises. That’s fine. The goal is to replace “whatever people found” with “what we’ve approved.”

Step 2: Sort your data into three buckets

Keep it simple:

  • Public: anything already on your website or in marketing materials. Fine to use with any AI tool.
  • Internal: procedures, internal emails, general business documents. Only in approved business AI tools.
  • Restricted: client or patient personal information, financial account numbers, Social Security numbers, health records, passwords, legal matters, HR files. Never in an AI tool unless it’s specifically approved for that data and you understand where it goes.

Post that list where people will see it. Three buckets are easy to remember.

Step 3: Pick approved tools and set them up properly

Choose one or two AI tools for the team and set them up through business accounts, not personal ones. Look for:

  • Sign-in with your company accounts, so access ends when someone leaves.
  • Clear terms on whether your data is used to train the vendor’s models.
  • Admin controls for history, sharing, and data retention.
  • Multi-factor authentication.

If you’re on Microsoft 365, it’s worth looking at AI options that run within your existing business account before adding another vendor.

Step 4: Write a one-page AI policy

It doesn’t need a lawyer’s touch to be useful. Cover:

  1. Which AI tools are approved, and how to request a new one.
  2. What data can and can’t go into them (your three buckets).
  3. That people review AI output before it goes to a client, especially numbers, names, and anything legal or financial.
  4. No AI browser extensions or plug-ins without IT approval.
  5. Who to tell if something sensitive was shared by mistake.

Step 5: Tighten the basics that make AI safer

AI doesn’t replace security fundamentals. It raises the stakes on them:

  • MFA on every account, including the AI tools themselves.
  • Clean up file permissions so AI assistants inside your business systems don’t surface sensitive files to the wrong people.
  • Fast offboarding, so former employees lose access to company AI accounts and their history the same day.
  • Keep software updated, including browsers, where many AI tools live.

Why California businesses should take this seriously

California has some of the most specific data rules in the country, and they apply no matter which tool caused the exposure.

Breach notification. California law requires a business to notify California residents whose unencrypted personal information was, or is reasonably believed to have been, acquired by an unauthorized person. According to the California Attorney General, a business that must notify more than 500 California residents because of a single breach must also submit a sample copy of the notice to the Attorney General.

The CCPA. The California Consumer Privacy Act applies to for-profit businesses that meet certain thresholds, such as gross annual revenue over $25 million, per the Attorney General’s CCPA overview. If it applies to you, it gives consumers the right to sue when their nonencrypted, nonredacted personal information is stolen in a breach caused by a failure to maintain reasonable security.

Even if the CCPA doesn’t apply to your business, your clients, insurers, and larger partners will increasingly ask how you control AI use. Having a policy and approved tools is a straightforward answer.

San Diego industries that handle sensitive data, like healthcare and dental practices, life sciences companies, law and accounting firms, and defense contractors, may also have sector-specific rules or contract requirements that restrict where data can go. Check those before approving any AI tool.

FAQ

Is it safe for employees to use free AI chatbots at work?
For public information, usually. For client, financial, health, or HR data, no, unless the tool is approved and set up through a business account with terms you’ve reviewed.

Do we need an AI policy if we’re a small business?
Yes. A one-page policy covering approved tools, what data can go in, and who to tell about mistakes prevents most problems.

What should we do if someone pasted sensitive data into an AI tool?
Tell IT right away. Delete the conversation if the tool allows it, review the tool’s data terms, and get advice on whether any notification obligations apply.

Related: Microsoft Copilot readiness · managed IT services

Get your AI use under control

Resolve IT 365 helps San Diego small businesses review which apps and AI tools are connected to their accounts, tighten Microsoft 365 permissions, and set sensible rules. Plans start at $249/month with unlimited remote support and no long-term contract. Request a free quote or call 858.331.6713.

Recent Posts

Categories

Sign Up for Our Newsletter

You can unsubscribe at any time, and we will NEVER sell your information to anyone.