Bei PDF Übersetzungen werden oft ganze Seiten nicht übersetzt, aufgrund einzelner kritischer Wörter.

Hi ​@Sisu_front_tier,

Thank you for raising this so clearly - you've actually put your finger on two real limitations of the current feature, and your feedback is genuinely useful for our team.

Why it doesn't show which words are flagged
The translation feature uses automatic safety checks that evaluate entire passages rather than individual words or phrases. The system only tells us that something in the text is unsafe - not exactly what triggered it. Because of that, and for safety reasons, the app currently can't highlight or identify the specific words involved. 

We know this isn’t as helpful as pointing to the exact word, and we’re actively looking at ways to make the experience less frustrating while still meeting our safety and legal obligations.

A workaround in the meantime
The best approach right now is to try translating smaller chunks - paragraph by paragraph, as you mentioned - to narrow down which section is causing the block. Once you've isolated it, you could translate the unaffected text with Acrobat and handle that specific passage manually or with another tool.

This isn’t meant to censor or change your original document. Your file stays exactly as it is. What’s happening is that the AI translation service is refusing to generate a translation for text that falls into those restricted categories. The filters apply in the same way to all users and all content, regardless of viewpoint, and are focused on reducing harm and complying with legal and platform requirements.

We recognize that this can affect legitimate use cases, like translating literature or historical documents that include sensitive themes. We’re continuously reviewing feedback and tuning the safety systems to better handle those cases while still meeting our safety and compliance obligations.

Rose