About NoLLMWM and its truth boundary
Text-watermark discussions often collapse invisible characters, statistical generation patterns, metadata, and AI-detection scores into one dramatic claim. We built NoLLMWM to separate those mechanisms in the interface. Physical Unicode can be inspected locally. Statistical vendor status remains unknown without the proper verifier. Rewriting can change language but cannot promise a private detector outcome.
That restraint is a product feature. It lets an editor make a useful change without being sold fictional certainty. Every result is worded so it can be repeated accurately to another person.
Privacy is implemented as less data
The easiest sensitive database to protect is one we do not create. Raw text stays local for scanning and is not persisted when rewritten. Logs and analytics accept safe identifiers and counts. Payment secrets and notification tokens remain server-only. Plans avoid unlimited language so operational limits are understandable.
We publish the data path, security controls, and limitations in plain English. As the product changes, those documents should change with it rather than becoming decorative promises.
Independent by design
NoLLMWM is not affiliated with Anthropic, Google, OpenAI, OpenRouter, or model vendors referenced in educational content. Their names help users understand distinct technical questions; they do not imply endorsement or official verification.
Questions about the product, responsible use, or accessibility are welcome through the contact page. For current provider behavior, consult the provider’s own documentation and terms.