Skip to content
Search
SecurityGlossary

Shadow AI

Shadow AI is the AI version of shadow IT. Someone signs up for a free chatbot, a browser extension or a meeting transcriber to get work done faster, and company data ends up in a service nobody reviewed. The risk is rarely the person's intent. The problem is that nobody knows where the data went, how long the provider keeps it, whether it's used for training, or who else can see it.

Banning AI outright can push use further out of sight. The defensive approach is to give people an approved tool that does the job, write a short AI acceptable use policy that says which data may go where, teach people to remove personal details before they paste (PII redaction), and keep an inventory of the AI services in use. The NIST AI Risk Management Framework gives a structure for that inventory and the governance around it, and guardrails cover the checks you can automate.

Example

A short shadow AI check-in for a small team (ask without blame):
1. Which AI tools did you use for work this month? Include browser extensions and phone apps.
2. What did you paste or upload into each one? (customer names, financials, contracts, code, none)
3. Was the account personal or company-managed?
4. Do the tool's settings allow your inputs to be used for training?
Then move the useful tools onto approved, company-managed accounts.

When to use it

  • Writing or updating your company's AI policy
  • Reviewing where customer or employee data flows
  • Onboarding new staff who already use AI tools on their own

When not to use it

  • Treating every use of AI as a threat; the goal is safe use, and people will use it either way
  • Relying on a ban with no approved alternative