Blog & Insights

What is shadow IT, and how is shadow AI different?

Shadow IT is software, services or outside help used for company work without IT knowing. Shadow AI is the same with AI tools, and it differs in ways that matter.

Shadow IT is any software, service or outside help that people use for company work without IT knowing about it or agreeing to it. Shadow AI is the same thing with AI tools. It differs because it takes minutes to start, because using it means handing it company data, and because it can now produce working software.

What counts as shadow IT

It is wider than most people think. The common forms are these.

  • Software bought without IT. A team signs up for a project tracker, a file-sharing service or a survey tool, on a free plan or a company card. Nobody reviewed the terms, and the account belongs to whoever created it.
  • A tool one department adopted. Sales runs its pipeline in one product, operations schedules in another, and marketing keeps its customer lists in a third. Each choice made sense to the team that made it. Company data now sits in places IT cannot see, back up, or close when someone leaves.
  • Spreadsheets and small databases doing a system’s job. A workbook that began as one person’s shortcut becomes the way quotes are priced or inventory is counted.
  • Consultants and contractors hired around IT. A department head brings in a firm to build an integration, a dashboard or an app. The work is connected to company systems before IT has seen the plan. When the engagement ends, nobody inside the company knows how it works.

None of this is misconduct. People had work to do and no quick way to get help, so they solved the problem with what they could reach.

Why it matters

What IT cannot see, it cannot protect or support.

  • Data. Customer records, contracts and financials are stored with vendors nobody vetted, under accounts nobody manages.
  • Access. Accounts stay open after an employee leaves, because IT never knew they existed.
  • Continuity. A process comes to depend on a product nobody is responsible for, or on the one person who set it up.
  • Cost. Several teams pay for tools that do the same job.
  • Obligations. Contracts and regulations apply to the data whichever tool holds it. If you handle regulated data, confirm with counsel what applies to tools adopted outside IT.

How shadow AI is different

  • It starts in minutes. There is nothing to buy or install. A browser tab and a personal account are enough.
  • Using it means giving it data. With most software, data goes in deliberately over time. With an AI assistant, pasting in the contract, the customer list or the source code is the use. What happens to that data depends on the vendor and the plan: how long it is kept, and whether it may be used to improve the vendor’s models. Read the terms for each tool.
  • It arrives inside software you already own. Vendors add AI features to products you approved years ago, and some are switched on by default. Google’s admin documentation, for example, says that the default setting for Gemini features in Workspace services is on.
  • It can build things. An employee can produce a working app with an AI coding tool and connect it to company data. That puts new software into the business without a review, and often without anyone who can maintain it. Before an AI-Built System Runs Your Business covers what to check.
  • Its output gets used. A wrong answer can reach a customer email, a contract or a report before anyone checks it.

How to find it

Ask before you audit. People will tell you what they use if the question is about helping them and they are not afraid of being punished for the answer.

  1. Ask each department head which tools their team would be stuck without.
  2. Read expense reports and card statements for software subscriptions.
  3. Review the sign-in logs of your identity provider for applications people reach with company accounts.
  4. Review the third-party applications that have been granted access to company email and file storage.
  5. Open the admin console of each major product you run and list the AI features that are turned on.

What to do with what you find

Sort each item into one of four groups: approve it as it is, approve it with changes (a company account, single sign-on, a limit on what data goes in), replace it with an approved tool that does the same job, or retire it. Tell the people who use it what you decided and why. For how the business and IT split the ownership of a tool once it is approved, see Who should own AI in a mid-market company, IT or the business?.

Then make the approved route the easy one. Shadow IT grows where asking IT is slower than signing up. A short written rule helps: the four-sentence AI policy is a place to start, and AI risk and governance covers the roles.

Common questions

Is shadow IT always a problem? No. Much of it is harmless, and some of it shows IT a need it had not met. It becomes a problem when it holds company data nobody is protecting, when a process depends on it and nobody supports it, or when it stays connected after the person who set it up has left.

Should we block AI tools until we have a policy? Blocking everything tends to move the use to personal phones and home computers, where IT sees nothing. Approving one or two tools on company accounts, with a short rule about what may go into them, gives people a safe way to do what they were going to do anyway.

Who should own the list of tools? IT keeps the list. Each tool on it should also have a named owner in the business who can say what it is for and whether it is still needed.

How often should we look for it? On a regular schedule, and whenever a department changes how it works, a new leader arrives, or the company makes an acquisition. Those are the moments when new tools appear.

If something you find is already connected to a system the business depends on, a Security Review of that system is the fixed-fee first step.

Related service: Security & Compliance

More like this, in your inbox. Subscribe for our latest thinking, resources, and events.

Get started

Not sure where you stand? Find out.