Built this while working on a larger B2B AI sovereignty project for European companies. The problem: most devs unknowingly send personal data to Claude/GPT/Mistral. This library detects and anonymizes PII locally before the API call — names, emails, phones, IBAN. Pure Go, zero deps, MIT license. Feedback welcome[](https://github.com/adminveez/Veez-pii-guard_
- _)
Top comments (1)
I also thought about this, and it makes sense. However, the main issue here is that, in general, it is not possible to create a 100% reliable solution. PII can be represented in different ways, and it’s also possible that some PII may not be detected and could be sent to an external LLM provider.