TrendAI Tops CyberGym Benchmark: 97% Success in AI Security

TrendAI's agentic exploit-remediation engine, codenamed AESIR, has achieved the top position on the CyberGym security benchmark. It scored a 97% success rate, surpassing all other current entrants on the leaderboard. This score is nearly 4 percentage points higher than the previous leader. CyberGym is an independent benchmark from the University of California, Berkeley. It evaluates AI security tools against over 1,500 confirmed vulnerabilities from 188 large open-source software projects. TrendAI uses seven AI models across four providers to maximize detection and remediation accuracy. The system aims to close exposure windows faster than attackers can exploit them. What's interesting is that TrendAI rapidly reduces risk by translating validated vulnerability intelligence into detection and virtual patches, without waiting for the normal patching cycle. This approach helps shrink the exposure window from weeks to as close to zero as possible. Sharda Tickoo, Country Manager for India and SAARC at TrendAI, emphasized that the goal is to make the gap between discovery and exploitation meaningless. The bottom line is that this technology aims to transform vulnerability management from reactive patching into proactive defense for enterprises.