Cloud.ru Introduces Guardrails Filter to Protect Data in LLMs

🛡 Cloud.ru Releases Guardrails Filter for Data Protection in LLMs

The tool works as a transparent proxy layer that masks personally identifiable information (PII), API keys, and passwords based on ~265 built-in rules (including Russian SNILS, INN, and OGRN) before sending the request to the model. Upon receiving a response, the tool automatically restores the original values.

🌍 The emergence of a fast proxy solution based on regex and validators allows companies to safely use public or private LLMs without transferring sensitive information to the provider. This lowers the barrier to AI adoption in regulated sectors.

👤 You can deploy this tool locally via Docker to protect your requests to ChatGPT, Claude, or GigaChat without changing your application logic — you only need to change the API base URL.

Source 1: https://github.com/cloud-ru-tech/guardrails-llm-filter