Understand how Clevera protects what's on your screen
What Clevera masks automatically when it reads your recordings, research, pictures and PDFs, what it never blurs or hides, and how to cover private details yourself before you share a video.
Screen recordings show whatever was on screen: inboxes, customer rows, keys in a settings page. Clevera takes a few automatic steps so private details don't get copied into your video, but it doesn't blur your footage for you. This article explains exactly what's protected, what isn't, and how to hide something yourself.
In short
| Clevera does | Clevera doesn't |
|---|---|
| Mask email addresses, card numbers, long numbers and keys in the notes the AI writes about your material | Blur or cover anything in the recording itself |
| Keep social-clip zooms from centering on private data it has spotted | Hide names, street addresses or other ordinary text |
| Fast-forward the time you held the Record with AI browser in how-to videos and product demos | Change what you or anyone else says out loud |
| Show password fields as dots, as every browser and app does | Check the finished video for you |
What Clevera does automatically
Private details are masked in the AI's notes
Before the AI uses your material, Clevera reads it and writes notes about it: what happens on screen, the text it shows and what is said. This covers screen recordings, uploaded videos, Record with AI recordings, website research, pictures and PDFs.
Before those notes are saved, anything with the shape of a private value is replaced with dots:
| What | How it's recognized | Replaced with |
|---|---|---|
| Email addresses | Text with an @ and a domain, including an address cut short in a narrow field |
•••@••• |
| Card numbers | Four groups of four digits, with or without spaces or dashes | ••••••••• |
| Long numbers | Nine or more digits in a row, such as an account number, an IBAN or a phone number written without spaces | ••••••••• |
| Keys and tokens | A run of 24 or more letters and digits with no spaces, such as an API key or a session ID | •••••••• |
Masking works by shape, not by words, so it works the same in every language. Prices, dates, times, version numbers and short order numbers keep their digits.
The AI director writes social clips from these notes, so the masked values aren't in the text it copies headlines, captions and voice-over lines from. It also looks at a few frames of your footage, so still check the finished clip.
Social-clip zooms steer away from private data
While reading your footage, the AI also marks areas that show someone's private data: email addresses, passwords and codes, keys, card or bank numbers, phone numbers, street addresses, names, and rows of customer data in a table, inbox or CRM. A product's own sample data isn't counted.
When a social clip zooms in on your footage and the camera would center on one of these areas, it moves off it or stays wider. This only affects where the camera points. The area can still be visible in the frame.
Your time with the Record with AI browser is fast-forwarded
When you take over the browser during Record with AI, for example to sign in, what you do stays in the recording. In how-to videos and product demos made from it, that stretch usually plays fast-forwarded, so your sign-in passes in a moment. It's still in the video, only faster. If you took the browser and never handed it back, that time is cut from the recording entirely.
Typing in the desktop app
The desktop app (Mac and Windows) records the keys you press so it can zoom to the field you type into, and uploads that keystroke log with the recording. The log itself isn't masked. The AI reads a summary of what you typed in which about a quarter of the characters are masked, so it knows that and where you typed without reading every character.
What you type also shows in the video wherever the app displays it. Pause the recording before you type a password, card number or other secret, and resume afterwards: nothing is captured while paused. See Understand what the desktop app captures.
What isn't masked
Be clear about the limits before you share a video:
- The picture itself. Nothing in your recording, screenshots, pictures or PDFs is blurred automatically. If it was on screen, it can be on screen in the video.
- Your keystroke log. The desktop app uploads the keys you press as they are; only the summary the AI reads is partly masked.
- Other kinds of text. Names, street addresses, company names and phone numbers written with spaces or dashes, such as
+1 555 010 0199, keep their text in the notes. - Audio. What you or anyone else says in a recording isn't bleeped or changed.
- Videos from your recording and articles. How-to videos, product demos, Polish videos and articles are written by the AI watching the recording directly, so the masking above isn't a guarantee there. Read the script and the article before you share them.
Hide something yourself
| To | Do this |
|---|---|
| Cover part of the screen | Add a Blur element over it: from the Elements tab in the video editor, or with + → Text & graphics → Blur in a social clip. A blur zooms with the recording, so it stays on the spot it hides. It covers only the time its block plays, and only in the language you added it in. See Add and edit elements. |
| Remove a moment | Leave it out of the footage: Edit sync and pacing in a how-to video or demo, Segments in a social clip. Or ask the assistant, for example "Don't show the inbox in scene 2". |
| Change words on screen or in the voice-over | Edit them by hand or ask the assistant. See Edit a social clip. |
Before you share a video
- Watch the whole video, at full size, before you export or share it.
- Look for email addresses, customer names, keys, card numbers and notifications in every scene, including the moments the camera passes quickly.
- Cover or cut anything private, then watch it again.