Hi! I'm David, co-founder of OpenShield. OpenShield is a transparent security layer that integrates your applications and AI models seamlessly. Building on traditional transparent proxies, it enhances them with cutting-edge AI techniques. Compatible with all significant LLM APIs, OpenShield allows you to apply rule-based policies and WAF or chain multiple LLMs to address issues like prompt injection, language forcing, and PII data classification. Simply replace the base URL in your API calls to benefit from robust, adaptable security tailored for AI-driven applications. Why do you need this? If you use AI in your product, it's important to know that it can create new security risks. You should take steps to protect your software, interfaces, and AI systems from outside attacks. Key features: Rate-limiting Caching Virtual keys Rule-based policies LLM model-based policies ( English language forcing, prompt injection, invisible characters) Coming soon: Rag based policies Hosted version Multiple AI provider support OpenShield is an open-source project. Please support us with a Github star!

Great to see OpenShield launching, David! The seamless integration with existing applications looks impressive. Excited to see how it enhances security for AI usage. Definitely giving this an upvote!