Shadow AI is any AI tool employees use for work without the company approving it: free chatbots, browser extensions, AI meeting note-takers, and AI features switched on inside other software. It matters because business information typed into those tools can be stored, shared, or used outside your control. The fix is not a ban. It is finding what is already in use, approving safe options, and setting a few clear rules.
In most small businesses, shadow AI is not a few rogue employees. It is nearly everyone, quietly using whatever helps them get through the day. That is good news: it shows where AI is already saving time. The goal is to keep that benefit and remove the risk.
Key takeaways
- Shadow AI includes more than chatbots. Browser extensions, meeting recorders, and AI apps connected to email or file storage are often the bigger risk.
- The most dangerous kind is an AI app that a user has given access to company email, calendars, or files, because it keeps that access after the browser tab is closed.
- You can find most of it in a few hours: ask staff, review connected apps in Microsoft 365 or Google Workspace, and check browser extensions and expense reports.
- Bans push AI onto personal phones. Approved tools plus a short policy work better.
What does shadow AI look like in a small business?
- Free chatbot accounts. An employee pastes a client proposal, a spreadsheet, or a difficult email into a personal ChatGPT, Gemini, or Claude account to rewrite or summarize it.
- Browser extensions. Writing assistants, summarizers, and "AI sidebars" that can read every page the employee opens, including the CRM, accounting system, and webmail.
- AI meeting note-takers. Bots that join calls, record them, and send transcripts to the vendor's cloud, sometimes to everyone on the invite.
- Connected AI apps. Tools a user has signed into with their work Microsoft or Google account and granted permission to read mail, calendars, or files.
- Built-in AI features. AI options inside software you already pay for, such as email, CRM, design, or document tools, that a vendor switched on by default.
- Personal phones. Photos of whiteboards, documents, or screens dropped into an AI app to transcribe or analyze.
Why is shadow AI a risk?
Data leaves your control. Consumer AI services may keep what users type and, depending on the provider and settings, use it to improve their models. Once a customer list or contract is in someone else's system, you cannot pull it back.
Connected apps keep the keys. When a user grants an AI app access to their mailbox or files, that app holds a token that keeps working until someone revokes it. If the vendor is breached, the attacker inherits that access. The ConnectWise SkyKick breach showed how a trusted vendor's app connection into Microsoft 365 can become a way in.
Confidential conversations get recorded. A note-taker on a call about a client's legal, medical, or financial situation can break your contracts or privacy obligations before anyone notices.
Compliance gaps. Healthcare, financial, and legal businesses must know where regulated data goes. An unapproved AI tool is a vendor you never assessed.
Wrong answers in client work. AI output that nobody checked can put errors in proposals, reports, and advice.
How do you find shadow AI in your business?
1. Ask, without blame
Send a short survey or ask at a team meeting: which AI tools do you use, for what, and what would you lose without them? Make it clear nobody is in trouble. You will learn more in one honest conversation than from any scan, and you will find out which uses deserve an approved tool.
2. Review apps connected to Microsoft 365 or Google Workspace
In Microsoft 365, the Entra admin center lists enterprise applications that users have consented to, along with the permissions each one holds. In Google Workspace, the admin console shows third-party apps with access to Google data. Look for AI tools, note-takers, and anything with read access to mail or files that you do not recognize.
3. Check browser extensions
If your computers are managed through Intune, Google admin, or another device management tool, pull a report of installed extensions. If they are not managed, check the computers of the people who handle the most sensitive information first.
4. Look at web and DNS logs
Your firewall, web filter, or DNS filtering service can show which AI sites people visit and how often. This tells you scale, not intent, so use it to plan approved tools rather than to single people out.
5. Check expenses and card statements
AI subscriptions often show up as small monthly charges on company cards or expense reports. They point you to the tools people value enough to pay for.
6. Review AI settings in the software you already pay for
Check the admin settings of your email, CRM, accounting, meeting, and document platforms for AI features, what data they can reach, and whether the vendor uses your data to train its models.
How do you bring shadow AI under control without banning it?
- Approve a short list. Pick one or two general AI tools on business plans with proper data terms, plus any specialist tools a team genuinely needs. Run each one through our AI tool vetting checklist.
- Write a one-page policy. Our AI acceptable use policy template covers approved tools, off-limits data, and how to request new ones.
- Require admin approval for new app connections. Both Microsoft 365 and Google Workspace can require an administrator to approve apps before users grant them access to company data.
- Remove risky connections. Revoke access for unknown or unneeded AI apps, and remove extensions that can read sensitive pages.
- Add data guardrails. Sensitivity labels and data loss prevention rules can warn or stop people before sensitive information leaves approved systems.
- Train with real examples. Show what safe and unsafe use looks like in your own business, and how to verify urgent requests now that voices and video can be faked.
- Check again every quarter. New tools appear constantly. A quarterly review of connected apps and extensions keeps the list current.
How Spot On Tech can help
Our AI governance service runs this process for businesses in New York and New Jersey: we inventory the AI tools already in use, review connected apps and permissions, help you choose and configure approved tools, write the policy, and train your staff. When we also manage your security, connected-app reviews become part of regular monitoring rather than a one-time project. Once the approved tools are in place, our AI implementation team helps staff use them well, so nobody needs to go looking for unapproved ones. For the bigger picture, read our guide to using AI safely in a growing business.
Frequently asked questions
What is shadow AI?
Shadow AI is the use of AI tools for work without the company's knowledge or approval, including free chatbots, browser extensions, AI meeting note-takers, connected AI apps, and AI features inside other software.
Is shadow AI the same as shadow IT?
It is a type of shadow IT. The difference is that AI tools usually receive the business information itself, such as documents, emails, and conversations, which raises the stakes for data privacy and confidentiality.
How can I tell if employees are using AI tools?
Ask them directly, then review connected apps in Microsoft 365 or Google Workspace, installed browser extensions, web filter or DNS logs, and expense reports. Together these find most of the AI in use within a few hours.
Should we block AI tools?
Block specific risky tools if needed, but do not block AI altogether. Bans push use onto personal phones where you have no visibility. Approve safe tools and publish a short policy instead.
Are AI meeting note-takers safe?
They can be, if they are approved, configured to store data in an account you control, and only used with everyone's knowledge. Unapproved note-takers on confidential calls are one of the most common shadow AI risks.
