Should you tell your boss you use AI at work?
You're not alone if you haven't. You might be more exposed than you think if you keep it that way.
Recent workplace research on what's now called shadow AI, employees using AI tools their employer hasn't approved or doesn't know about, found that more than half of workers use personal AI tools without approval, and 66 percent of office professionals at large companies use AI tools they believe aren't permitted under company policy. This isn't a fringe behavior. It's closer to the default.
Why people don't disclose
Mostly not deception. Usually it's some mix of not wanting to seem like they need help, not being sure the tool is technically banned versus just undiscussed, and reasonably assuming that everyone else is doing it too, so it isn't worth a conversation.
Where the actual risk sits
The risk isn't really about getting caught using a tool. It's about what you put into it. Workplace data shows employees regularly input sensitive information, internal messages, HR data, confidential documents, into unapproved AI tools, and under regulations like GDPR and HIPAA, that can constitute a reportable violation regardless of intent. With EU AI Act enforcement landing in August 2026, this is becoming a live compliance issue for employers, not a theoretical one, and that pressure tends to flow downhill to individual employees fast once a company starts paying attention.
A reasonable way to think about disclosure
Separate two questions that often get conflated: "should I use AI for my work" and "should I put this specific piece of information into an AI tool." The first is increasingly just how modern work gets done. The second deserves a real pause every time, especially for anything involving client data, personal information about coworkers, or anything under a confidentiality agreement.
If your use is genuinely just drafting, summarizing your own notes, brainstorming, low sensitivity, disclosure is more about transparency and less about risk. If it involves company or client data of any kind, that's a different conversation, and one worth having proactively rather than waiting for a policy to force it.
How to raise it without it sounding like a confession
Frame it as a process question, not an admission. Something like asking your manager or IT team what tools are actually approved, rather than announcing what you've already been doing. Most workplaces are behind on writing formal AI policy, and asking the question often prompts them to write one, which benefits everyone, not just you.
Take the career-risk assessment
Frequently asked questions
Can I get fired for using an unapproved AI tool?
It depends heavily on your employer's specific policy and what data you put into the tool. Using AI for your own drafting is rarely a firing offense on its own; putting confidential data into an unapproved tool is a more serious matter in most workplaces.
What if my company has no AI policy at all?
That's extremely common right now. Asking what's approved, rather than assuming silence means permission, is the safer read given how fast enforcement expectations are shifting in 2026.
← Back to the assessment · Related: What HR actually thinks about AI on your resume