A study by Inetum among Belgian executives reveals a striking figure: 68% of them already use artificial intelligence at work. Nearly half do so without telling their own management, and that share rises to 55% in SMEs with 10 to 250 employees. This phenomenon has a name: shadow AI, the unmanaged use of artificial intelligence inside a company. It is not a problem limited to large groups, nor to employees alone.
Mass adoption, driven from the top
Artificial intelligence has taken hold quickly in Belgian companies. According to Inetum, 68% of executives now use it, peaking at 76% in companies with more than 250 employees. Executives under 45 are especially active users: 30% of them use it daily. The perceived benefits are real: 75% report a productivity gain, and 64% report improved working conditions. A quarter of users, however, notice no impact at all, a sign that AI is still often confined to simple tasks rather than strategic ones.
The real problem: AI without a framework, even at the top
The most revealing figure is not about adoption itself, but about how it happens. 47% of Belgian executives use AI without their own management knowing, and 18% even rely on tools that have not been approved, a share that climbs to 27% among executives under 45. In mid-sized companies, with 10 to 250 employees, 55% of executives use AI in secret. Shadow AI is therefore not primarily a discipline problem among staff: it often starts at the top of the organisation, for lack of a clear policy.
Concrete risks for your business
Pasting an internal document into a consumer AI tool can expose sensitive information: customer data, financial figures, business strategy, source code or confidential HR data. These tools do not necessarily comply with GDPR, and some data may be stored outside Europe or kept longer than necessary. The logical consequence: six in ten Belgian executives now say they are concerned about the risk of cyberattacks, a risk that unmanaged AI use only amplifies.
The AI view, humans first
Shadow AI is not proof that artificial intelligence is too risky for business. It is a symptom of AI arriving before the company set its own rules. Banning it outright would be a mistake, just as turning a blind eye to it would be: 75% of executives who use it already report real productivity gains. The right approach is to secure the tools, clearly limit access rights, and keep human oversight over AI-generated output, especially for strategic decisions. AI executes. Expertise decides, and watches over it.
This week's to-do
Take stock, even informally, of the AI tools used by your marketing, sales or HR teams. Check that multi-factor authentication is enabled on exposed accounts. Write, in one page, what is allowed, managed or forbidden when it comes to AI in your company, and share it clearly with your teams.
Setting up properly managed AI agents and assistants, with defined access rights and clear oversight, falls under Vistalaro Build: we frame automation before it becomes a risk. Defining your company's AI usage policy and training your teams is part of the support provided by Vistalaro Pilot, your dedicated senior marketer.
Is AI in your business really under control?
We review your AI usage together, no jargon, and put guardrails where they are needed.
Let's talk