Shadow AI: why workers bring their own AI, and what to do about it
Gartner found 88% of workers with enterprise AI also use personal AI tools. Here is why shadow AI happens, the real risks, and a six-step plan to fix it.
Shadow AI is the use of AI tools at work that the company has not approved, usually through personal accounts. Most workers do it even when they have an approved tool: Gartner found that 88% of employees with access to enterprise AI also use personal AI tools for work (reported by CloudNews). Banning it rarely works. What works is understanding why people do it, setting clear data rules, and making the approved path the better one.
What shadow AI looks like day to day
It is rarely dramatic:
A marketer pastes customer interview notes into a personal ChatGPT account to pull out themes.
An engineer uses a personal Claude account because the company tool has a different model.
A sales lead drafts a proposal in Gemini on their phone between meetings.
A finance analyst uploads a spreadsheet to a free AI tool to clean up formulas.
None of these people think they are doing anything wrong. They are trying to get work done faster.
BetterCloud's 2026 State of SaaS report puts it plainly: "shadow AI is the new shadow IT." It found that only 56% of applications in use carry IT approval, and 18% of organizations identified data leaks originating from AI tools and chatbots in the past year (BetterCloud).
Why people bring their own AI
If 88% of people with an approved tool still use a personal one, access is not the issue. These are the usual reasons.
The personal tool knows them
Personal accounts build up history. ChatGPT can keep saved memories and reference past chats (OpenAI Memory FAQ). After months of use, a personal account has context the fresh company account lacks. Switching feels like starting over. We explain why in why every AI chat starts from zero.
The approved tool is weaker for their task
Different tools are better at different jobs. If the company picked one tool and someone's work suits another, they will use the other.
The approved tool lacks company context
The approved tool is often allowed to see company data but has not been given any. So it gives generic answers, and people go back to the tool they know.
Nobody explained the rules
Many teams roll out a tool with a long policy nobody reads. People are not sure what is allowed, so they do what is fastest.
The real risks
Shadow AI is not only a security issue. There are three separate risks.
Risk | What happens | Example |
|---|---|---|
Data exposure | Company data goes into a tool with consumer data terms | Customer contracts uploaded to a free account |
Lost work | Useful prompts, outputs and decisions stay in personal accounts | A strong analysis leaves when the person leaves |
Inconsistent answers | Each person's AI works from different context | Two reps quote different discount rules |
The data terms differ between consumer and business plans. Anthropic trains new models on data from Claude Free, Pro and Max accounts when the user's training setting is on, while its Team and Enterprise plans fall under commercial terms that this does not apply to (Anthropic). OpenAI says it does not train on data from ChatGPT Business, Enterprise or its API platform by default (OpenAI). Check the current terms for any tool your team uses.
The lost-work risk gets less attention, but it compounds. Every month, more of your team's thinking ends up in accounts the company cannot see.
What to do about it: a six-step plan
1. Ask before you act
Run a short, anonymous survey. Ask three questions: which AI tools do you use for work, what for, and why that tool? You will learn more from the "why" than from any network scan.
2. Set data rules by type, not by tool
People remember categories better than tool lists. A simple version:
Public: published content, public docs. Any tool.
Internal: plans, internal docs, non-sensitive metrics. Approved business tools only.
Confidential: customer data, contracts, financials, personnel. Approved tools with business data terms, and only when needed.
Restricted: credentials, health data, anything under legal hold. Never in an AI tool.
3. Give people a tool that is at least as good
If the approved tool is worse than the personal one, people will keep using the personal one. Ask the survey respondents what would make them switch. Sometimes the answer is a second approved tool, not a better policy.
4. Put company context into the approved tool
This is the strongest pull you have. Set up shared projects with your product docs, brand voice, pricing rules and recent decisions. An approved tool that knows your business beats a personal tool that does not.
5. Write a one-page policy
Keep it short enough to read in two minutes. Our guide on how to write a team AI policy has a template.
6. Bring good work back
Make it normal to save useful outputs, prompts and decisions to a shared place. When someone finds a prompt that works, turn it into something the whole team can reuse.
Don't treat it only as a problem
Gartner's same research found that employees who combine personal and corporate tools are 1.7 times more likely to report significant time savings (CloudNews). Shadow AI is often a sign of your most engaged people. They have already found which tools and workflows help. Your job is to learn from them, then move that value into a place the whole team can use safely.
FAQ
What is shadow AI?
Shadow AI is any use of AI tools for work that the company has not approved or cannot see. It usually means personal accounts on public chat assistants.
Is shadow AI illegal?
Not by itself. The risk is what data goes in. Customer data, personal data or confidential contracts in a consumer tool may break your contracts, your privacy commitments or data protection law.
Should we block personal AI tools?
Blocking alone tends to push use onto personal devices, where you have even less visibility. Pair any blocks with an approved tool that is good enough to replace what people were using.
How do we find out what people use?
Ask them. An anonymous survey with a clear promise of no penalties gets honest answers. Then check SaaS spend and single sign-on logs to fill the gaps.
The short version
Most workers bring their own AI because their personal tool knows them and the approved one does not. Treat that as information. Set simple data rules, give people a tool that is at least as good, load it with real company context, and make it easy to bring good work back into the open. For the wider cost of running many tools, see the hidden cost of AI tool sprawl.
We're building Shane to give every approved AI tool your team's context, with sources and permissions, so the sanctioned path is also the better one. If that's a problem your team has, you can join the waitlist at tryshane.ai.