A security manager at an insurer wants to know which AI tools are being used across the organisation and asks IT for an inventory. The same day, HR replies: this cannot happen without the works council, because it means tracking employees. Two emails, and the project stalls.
Both are right about their own interest, and both assume one comes at the expense of the other. That is where they go wrong. The organisation wants to know which tools are in use; the employee does not want a record of what they do. Answering the first question requires no names, so the second concern need never arise.
What does the organisation actually want to know?
Shadow AI is AI use people have taken up on their own: a tool anyone can start using online, that the organisation has not decided on and that is actually in use. Our introduction to shadow AI explains what that means and why a ban does not solve it. The security manager's question is the natural next step: which of those tools do we have, and which ones need a decision?
That is a question about services. The answer is a list of domains, with a count and a status for each: allowed, limited use, no decision yet, not allowed. There is no name anywhere in that list, and none is needed. The insurer remains the controller for what happens inside those tools, so the interest is real. It just does not require what HR fears.
What does the employee actually want to avoid?
It is not that the organisation should see nothing at all. Most people understand that an employer wants to know whether client data is ending up in a free chatbot. What they do not want is a record of themselves: which tools they use, how often and at what times, and the chance that this comes up again in an appraisal or a dispute.
That is a reasonable concern, and the GDPR takes it seriously: processing employees' data requires a purpose, necessity and transparency in advance, and in the Netherlands the Works Councils Act adds its own requirements. What an employer may and may not do is covered in our article on whether your employer can monitor your AI use. But note what the concern is about: the record of the person, not the count of the tool.
Where is the difference?
In one field. An observation without a user ID says: on this day, this domain was opened in the organisation. An observation with a user ID says: this person opened this domain. The first is a count, the second a behavioural record. They can come from the same source, yet they are two different things, with a different legal basis, a different retention period and a different conversation with the works council.
| Interest | What it asks for | What it does not ask for |
|---|---|---|
| Organisation: a view of shadow AI | Which domains, how often, which status, how that changes | Who opened the tool |
| Organisation: getting the decision across | Showing the decision when someone opens the tool | Recording who saw the notice |
| Employee: no record | No user ID, no content, no breakdown per person | That the organisation sees nothing |
| Employee: knowing where they stand | The decision per tool, and the alternative | A ban without explanation |
The top two rows and the bottom two fit together. Not because someone struck a compromise, but because the organisation's question was never about a person.
Why does monitoring individuals also work against the organisation?
Suppose the insurer decides to measure per employee after all. What happens? The 2025 study by KPMG and the University of Melbourne of 48,000 workers in 47 countries found that 57 per cent of workers hide their AI use from their employer. That is the starting point, before anything has been announced.
Announce that use will be visible per person, and it moves to personal phones and personal accounts. The security manager's figures drop, the use does not and the visibility the whole project was meant to deliver is gone. Monitoring individuals costs the organisation precisely the insight it was introduced to provide. A count without people has no such effect, because there is nothing to evade.
What does this look like in practice?
The insurer sets up a measurement based only on domain and count, presents the design to the works council and switches it on. After six weeks, the picture is this: Copilot is allowed and is opened most often. ChatGPT and Perplexity also show up daily, along with three specialist tools nobody knew about: one that summarises policy terms, a translation tool and a transcription tool. Two of them have no decision yet, because they were not on any list.
No incident, but work to do. The summarising tool sees clients' policy documents, so it gets a decision first. The transcription tool follows. Nobody needs to know who used them to decide that, and the works council saw the list before the security manager did. It also shows what a one-off inventory misses: nobody could have asked about the three unknown tools, because nobody knew they existed.
What does BeeSensible record?
The extension notices when someone opens an AI tool and reports only the registrable domain, at most once a day per device, with no user ID, no path and no page content. Every tool starts as "No decision yet"; the organisation sets the status itself. When someone opens a tool that has a decision, that decision appears in the browser, with the approved alternative as the first button; the organisation decides whether people can still go on to the tool. What happened after such a notice is counted per tool, without a user ID.
