Screenshots: When Smart Blur Blurs the Whole Screen

Even on a project set to Smart Blur (AI), SCREENish sometimes blurs the whole screenshot instead of just the sensitive parts. This is automatic, you do not need to do anything, and it always errs on the safe side: whole-screen blur can never reveal more than smart blur would.

• The screen looks like a configuration or source-code file. When many credential-style labels appear on one screen (passwords, API keys, tokens, database connection strings — seven or more), these are the hardest screens to redact precisely, so the whole screen is blurred to be certain nothing leaks.
• The screen is very text-dense. A packed code editor or a long document (more than about 3,000 recognised words) holds more text than can be reliably checked, so the whole screenshot is blurred rather than risk leaving a name visible.
• The computer is low on memory. Smart Blur needs working memory to read the screen. Under memory pressure the app skips the heavier smart pass and blurs everything instead.
• Redaction is taking too long. On slower machines, if analysing a screenshot would take longer than about two minutes, a whole-screen blur ships instead so uploads never fall behind.
• The on-screen text cannot be read. If text recognition is not ready yet or fails for any reason, the app falls back to whole-screen blur so a screenshot is never uploaded without redaction.

In every case your activity stays visible — layout, colours and the general shape of the work — but no readable text is exposed. With blur enabled, SCREENish never uploads an un-blurred screenshot. To pick a mode in the first place, see Screenshots: Choosing a Privacy Blur Mode.