Shadow AI is already in your company

When we ask leadership teams whether their staff use AI assistants at work, the answer is often "not officially". When we ask the staff, nearly everyone does. They summarise documents, draft emails, debug code and analyse spreadsheets, often with free consumer accounts and personal logins.

That gap is the real risk. Not that people use AI, but that they do it in tools the company has never reviewed, with data the company has never classified, and with no record of what was shared.

What actually goes wrong

• Sensitive data in consumer tools. Customer records, unreleased financials or source code pasted into a service whose terms allow it to retain inputs or use them to improve its models.
• Unvetted integrations. Browser extensions and plugins that request permission to read every page or mailbox a user opens.
• Overconfident output. Generated code, contract language or analysis accepted without review, then shipped to customers.

Give people a good approved option

Choose one or two business-grade AI tools and make them easy to access. Before you approve them, check:

1. Whether your inputs are used for training, and whether you can turn that off contractually.
2. Where data is stored and for how long, and whether that fits your data protection obligations.
3. Single sign-on, admin controls and audit logs, so access ends when someone leaves.

Then tell everyone clearly that this is the tool to use.

Write a one-page policy people will read

Long policies get ignored. A useful AI policy fits on one page and answers four questions:

• Which tools are approved, and for what kinds of work.
• What data must never go in, such as personal data about customers, credentials, or information under NDA, unless the tool is explicitly approved for it.
• Who checks the output. A human stays responsible for anything that goes to a customer, into production code or into a legal document.
• How to ask for something new, so a request for a new tool gets an answer in days, not months.

Get visibility without spying

You can learn a lot from tools you already have. Your identity provider shows which AI apps people sign in to with their work accounts. Browser and endpoint management can list installed extensions. Data loss prevention features can flag large pastes of sensitive data into known AI domains. Use this to shape the approved list, not to punish people.

Review it every quarter

AI tools change faster than almost any other category of software. Features appear, terms of service change, and new risks such as AI agents that act on a user's behalf arrive quickly. Put a short review on the calendar every quarter.

The companies doing this well are not the ones with the strictest rules. They are the ones where using AI safely is the easiest option available.