Frequently asked questions

The extension checks the text in the input fields of the apps where BeeSensible helps. By default, that text goes to BeeSensible's own servers, hosted by European providers, where it is checked in working memory and discarded straight afterwards. BeeSensible keeps only counts: which type of data was found, in which app and what happened to it.

BeeSensible does not block sending. While you type, sensitive data is highlighted, and you choose whether to mask it, remove it or leave it. If your organisation has turned on Mask on send, anything still highlighted when you send is replaced with a placeholder such as [NAME1]. The final check then holds the message for a couple of seconds at most, and if that check is not back in time, the message is sent as it is. Policy > Masking explains this in more detail.

That depends on your organisation. By default the notice shows the allowed alternative, and you can still go ahead with the "Hold to continue anyway" button. If your organisation chooses Employee cannot continue, the notice only shows the alternative and the Go back button. See Notices for AI tools.

Detection relies on models and patterns, so text that looks like sensitive data can also be highlighted. If the same unnecessary highlight keeps coming back, an admin can set that data type to Standard or switch it off in the detection profile.

An admin cannot see what anyone has typed. The dashboard shows counts per data type, app and action, but never the content. There is also no view per employee, and for detections, documents and visits to AI tools BeeSensible does not record which user they came from.

Usage figures only appear once your organisation has five active users, meaning members who have signed in to the extension or the desktop app. Which AI tools have been seen, and which are new, does stay visible at all times, so the Inbox works from the very first visit. See Organisation settings.

You only register the tools you allow. A tool you allow, or allow with limits, in the Inbox goes into the AI register automatically. There you add the use cases, the responsible people, the documents and a review date. See The AI register.

As long as you make no decision, nothing changes for the tool. It stays on No decision yet and remains in the Inbox while it is in use, and by default employees notice nothing. Admins receive one email each morning when new tools have come in. See Inbox: tools without a decision.

The risk score is not a legal judgement. The score across six dimensions and the estimated classification under the EU AI Act are BeeSensible's assessment and not legal advice. You assess the tool and make the decision yourselves. See Catalog and risk.

Accountability is not a compliance statement. The page shows what you can demonstrate with the measurements you already have, and explains for each legal provision what the law asks, what BeeSensible contributes and what falls outside it. It shows no compliance status anywhere, because the assessment stays with the controller. See Accountability.

Ask your colleague to check their spam folder first. If the email is not there, choose Resend invitation for that member in the members list under Organisation.

Click the puzzle-piece icon in the Chrome or Edge toolbar and pin BeeSensible. The icon then stays next to the address bar.

Document anonymisation is a separate module of your subscription. When that module is off, Documents does not appear in the dashboard, the extension or the desktop app. To add the module, email hello@beesensible.eu.

Besides PDFs, you can by default also anonymise Word, Excel, PowerPoint, OpenDocument, RTF, plain text, Markdown, CSV, scans and images, and every file comes back in the same format. Your admin can switch the extra formats off under Policy > Masking (More file formats for documents). The desktop app works with PDF only. See Anonymising documents.

You configure this under Policy > Detection profiles. For each profile you choose which data types are highlighted and at which level (Standard is yellow, Critical is red), and under Apps and documents what the profile applies to. On the Custom terms tab you add terms that are sensitive only for your organisation, such as project names, client names and codes. See Detection profiles.

BeeSensible keeps the counts behind the dashboards for 24 months at most, and a daily clean-up removes anything older. It never stores typed text. The AI register and the decisions are kept for as long as your organisation exists.