TrendAI Tops CyberGym AI Security Benchmark with 97% Success

TrendAI's exploit-remediation engine, codename AESIR, has achieved the top position on the CyberGym AI security benchmark. It reached a 97% success rate, placing it ahead of all other entrants. This score is nearly 4 percentage points higher than the previous leader. CyberGym, a benchmark from the University of California, Berkeley, evaluates AI security tools against over 1,500 confirmed vulnerabilities. These vulnerabilities come from 188 large open-source software projects. Rachel Jin, Head of TrendAI, states that this validation changes what is possible for enterprises adopting AI. She emphasizes the goal of compressing the exposure time of vulnerabilities to nearly zero. Mick McCluney, ANZ Field Chief Technology Officer for TrendAI, highlights the company's use of seven AI models from providers like Anthropic, Google, and OpenAI. This enhances their ability to maximize detection and remediation accuracy. The TrendAI approach identifies weak spots, validates them, and then rapidly reduces risk by creating detection and virtual patches. This process aims to shrink the exposure window from weeks to as close to zero as possible. This matters because it offers a proactive defense against fast-moving threats.