Show HN: AI Watermarkremover (aiwatermarkremover.online)

AI Watermarkremover is a simple web tool that scans pasted text from ChatGPT, Claude, Bard and other assistants for hidden unicode markers—like zero-width spaces, zero-width joiners and other invisible characters—and strips them while preserving visible content. It also offers cleanup options such as handling em dashes (replace or remove) and claims coverage for a broad range of invisible-character watermarking techniques. The site frames this as a way to eliminate “invisible” watermark payloads that some providers have discussed embedding in model outputs. For the AI/ML community this highlights two practical and ethical points. Technically, the tool targets explicit, character-level watermarking schemes (invisible unicode characters), which are easily removed once the patterns are known; it does not defeat statistical or linguistic watermarks that alter token selection distributions. That distinction matters because character-based and probabilistic watermarking are fundamentally different arms of the provenance problem. Ethically and legally, removing invisible characters is generally permissible for personal use, but it can undermine content attribution and detection systems—so users must still heed platform Terms of Service and disclosure requirements. In short: useful for cleaning accidental or deliberate invisible characters, but not a catch-all solution to provenance or detector-evasion debates.