AIAdopt
whitepaperFebruary 2026· 7 min reading time

Shadow AI in SMEs: risks, scale and how to respond

What shadow AI is and the risks it brings for data, quality, compliance and reputation. With an approach to channel it instead of just banning it.

Shadow IT is a familiar phenomenon: employees using their own tools and systems without IT's approval. Shadow AI is the new variant. It is growing faster than an IT department can keep up with.

This whitepaper describes what shadow AI is, how widespread it is, which risks it brings and how to deal with it as an organisation without stifling innovation.

What is shadow AI?

Shadow AI is the use of AI tools by employees without the organisation knowing about it or having approved it. It is not about malice. It is about employees who want to do their work better or faster and use tools they have found themselves.

Concrete examples: an employee who has customer emails rewritten by ChatGPT. An HR employee who has application letters summarised by an AI tool. A marketing employee who has images made by a free AI service. A finance employee who uses an AI browser extension to analyse spreadsheets.

All these situations have two things in common. The organisation does not know about them. And company data is processed through channels that the organisation has not approved, assessed or covered by a contract.

The scale of the problem

The use of generative AI at work has grown rapidly. A substantial share of it takes place outside the employer's approval, policy or view. In the Microsoft and LinkedIn Work Trend Index of May 2024, three quarters of knowledge workers said they use AI at work. Of those users, 78 percent brought their own AI tools to work, so-called "bring your own AI". This happened in every age group, most often among younger employees.

At small and medium-sized companies, that share was slightly higher still, at 80 percent. Microsoft does use its own classification by company size for this. Shadow AI is also particularly hard to control in SMEs. Large companies more often have security teams and software that spot traffic to public AI services. That is precisely what SMEs often lack.

The result is that a smaller organisation often has little visibility into which AI tools employees use, which data they share with them and which risks that brings.

The risks

Shadow AI brings four kinds of risk.

1. Data loss and privacy

When employees enter company data into public AI tools, that data leaves the organisation. It is processed on external servers, possibly outside the EU. Depending on the tool, the type of account, the settings and the contract, it may also be used to improve AI models.

Browser extensions are a particularly risky form of shadow AI when they have broad permissions. An employee thinks they are using a handy summarise button. Meanwhile the extension may be able to read the text on the screen, including confidential emails, financial data and customer information. In effect, it can become a data breach or unlawful processing, with an AI veneer.

This is not only a business risk. It is also a risk under the GDPR, the European privacy law. In Belgium, the Data Protection Authority can become involved where shadow AI leads to unlawful processing of personal data or to a data breach that has to be reported.

2. Quality risks

AI tools produce text that sounds convincing but can be factually wrong. These are called hallucinations. When employees without training use AI-generated text without checking it, incorrect information can end up in customer communication, reports or decisions.

A concrete example: lawyers have been sanctioned and fined for including case law invented by AI in court documents. The AI had simply made those cases up. Nobody had checked before the document was filed.

3. Compliance risks

To comply with the EU AI Act, you as an organisation need to know in practice which AI systems you use and which risk level applies to them. For high-risk AI, the law also sets specific requirements for documentation and use. If you do not know that employees are using AI, you cannot map it. And if a supervisory authority asks which AI systems you deploy, "we do not know" is not a strong answer. What the law asks of you on AI literacy is set out on the EU AI Act page.

The AI Act also prohibits a number of practices that employees could use without realising it. Emotion recognition of employees in the workplace? Prohibited, except for medical or safety reasons. Using AI to gauge customers' moods is not automatically prohibited. It does touch on transparency, the GDPR and consumer protection. Giving people a score for their behaviour (social scoring) and disadvantaging them unjustifiably or disproportionately as a result? Prohibited. Other forms of scoring may be high-risk rather than prohibited, depending on purpose and effect. We set out separately when you do have to tell people that AI is involved. Our free EU AI Act Compliance Checklist helps you work out in ten steps where your organisation stands.

4. Reputational risks

Imagine a customer discovering that their confidential data was processed by a public AI tool. Or an applicant discovering that ChatGPT summarised their cover letter without their consent. The reputational damage can be greater than the legal consequences.

Why banning alone does not work

The first response to shadow AI is often: ban it. Block ChatGPT on the company network. Ban AI browser extensions. Write a strict policy with sanctions.

Banning alone does not work. Blocking risky tools can be a necessary part of your policy. But a ban without approved alternatives, policy and training usually pushes use further underground. Employees often find a way around it. Shadow AI is almost always a signal of need and not of bad intent. So do not see it as rebellion. See it as free market research into what your own team needs. And AI is not going away: organisations that ban AI run the risk of falling behind. Often people also simply get too little out of the tools they find on their own.

The approach: channel instead of just blocking

•Step 1: Map current use. Ask employees openly and without threat which AI tools they use. For the use of AI chatbots, you can ask that anonymously with the free AI Adoption Scan. Also check network traffic and installed browser extensions. Would you rather have the whole inventory done for you? That is what an AI Adoption Assessment is for.
•Step 2: Offer approved alternatives. A business licence for ChatGPT or Copilot, for example, with the right settings and contractual terms so that your data is not used for training. That is a real difference from the free versions.
•Step 3: Draw up a policy. An acceptable use policy for AI. Short, clear and concrete. Discuss it in a team meeting and do not just send it as an attachment to an email. If you do not have a policy yet, you can start with our two-page template. Not sure whether you need one? Run through the five signs.
•Step 4: Train your staff. Employees who understand why shadow AI is risky often stop of their own accord if they have a good alternative. Training is not the stick. Training is the carrot. Our AI literacy training under the EU AI Act is built for that. Anyone who mainly wants to get more out of AI is in the right place with Effective AI at Work. What that costs per person is stated openly on the pricing page.

The role of the IT department

IT can block tools, monitor network traffic and manage business licences. But the policy must come from leadership, the training from HR and the culture change from everyone. Whether your organisation counts as a provider or a deployer changes what is expected of you. A provider develops an AI system. A deployer uses it.

A smart approach: use your firewall not only to block but also to redirect. If an employee browses to an AI site that has not been approved, send them to an internal page with the message: "Want to use AI? Good idea. Use our secure business environment and complete the AI literacy training first."

Conclusion

You do not solve shadow AI by banning it alone. It is a signal that your organisation is lagging behind what employees need. The solution is not less AI but better AI: approved, secured and supported by policy and training.

If you take shadow AI seriously, you often solve two problems at once. The compliance risk becomes smaller. And productivity can rise because employees work with better tools, in a safer way. We calculated separately what training delivers.

Need help tackling shadow AI in your organisation? AIAdopt supports you through the process, from inventory and risk assessment to policy and training. In our AI literacy training, shadow AI comes up in the modules for employees, managers and IT. The free AI Adoption Scan anonymously maps which AI chatbots your employees are using on their own. If you would rather have the inventory and risk assessment done for you, that is what an AI Adoption Assessment is for. Or you can simply get in touch.

Source: Microsoft and LinkedIn, AI at Work Is Here. Now Comes the Hard Part, Work Trend Index 2024, published 8 May 2024.


Written by Rob Ummels in collaboration with Claude (Anthropic). Final editorial responsibility: Rob Ummels.

Want to know where your organisation stands?

Download our free EU AI Act Compliance Checklist or view our AI literacy training.

You'll want to know this too...

All insights →
🧩
September 2026

AI governance: what to put in place and how it connects

AI governance explained with an infographic and clear guidance. Discover how direction, data, people and oversight connect, including in small organisations.

Read more →
🔍
October 2026

Many organisations do something about AI literacy. Who checks whether it works?

Many organisations do something about AI literacy. But who checks whether it works? Find out how to check what employees know and actually apply in their work.

Read more →
🏛️
September 2026

What is AI literacy? The Council of Europe's answer

Article 4 does not say how far AI literacy has to go. The Council of Europe does describe it, in three sides. Read what that means for you in practice.

Read more →

Shall we talk?

Get in touch for a no-obligation conversation about what AIAdopt can do for your organisation.