Recon by Protect AI is an advanced product for identifying potential vulnerabilities in Large Language Models (LLMs), ensuring enterprises can deploy AI applications with confidence. Recon runs automated penetration testing on LLM powered applications to protect against unique security threats and vulnerabilities, enabling AppSec and product teams to deploy GenAI applications quickly and confidently. It uses an attack library and an LLM agent-based solution for red teaming and evaluating the security of GenAI systems. Recon evaluates the security and safety of GenAI systems using an attack library and an LLM agent-based solution. Automated Red Teaming uses a comprehensive attack library to run detailed automated LLM attacks from categorized threat profiles, including jailbreaks, prompt injection attacks, and input manipulations, crucial for maintaining AI systems' integrity and security. Recon includes a library of 20K+ known vulnerabilities for attacking GenAI systems. The library updates weekly with new techniques and allows adding your own threat research. No-Code Integration enables scanning custom endpoints for vulnerabilities in under 5 minutes, with automated scans running asynchronously and notifications upon completion. Recon provides insights into attacks needed to compromise your system, reducing red team effort to test your GenAI system. It continuously improves LLM security with automated red team tests to identify vulnerabilities in LLM systems.
