Anonymous-by-Construction: An LLM-Driven Framework for Privacy-Preserving Text

The project addresses a common enterprise need (PII redaction) using current LLM-based techniques rather than traditional regex or NER. While 'type-consistent surrogates' (replacing a name with another name rather than a tag) improve data utility, the approach is a standard application of local LLM prompting. With 0 stars, it currently lacks the adoption or unique architectural moat required for higher defensibility. Frontier labs and cloud providers (AWS, Azure) are rapidly integrating similar PII-masking layers into their primary offerings.