AI Text Anonymizer

This is a first pass, not a compliance tool. In our own testing a small model reliably caught most names, emails, and phone numbers but still left an occasional one in place, and sometimes listed an entity in the mapping while missing it in the text, so the mapping is a checklist to verify against, never proof of coverage. Read every line of the output yourself and search it for the specific names you know are in there before sharing it. It pairs well with the regex-based PII Scrubber, which never misses a pattern it knows but cannot spot contextual identifiers like nicknames or project code names; running both catches more than either alone. For regulated data with a legal obligation, use a validated redaction process. Powered by the same private engine as the Browser SLM Lab.