< Back to all clusters
[TECHNOLOGY] · United States · 6 sources

started · updated

TrendAI ranks first on CyberGym AI security benchmark

TrendAI's agentic exploit-remediation engine, codenamed AESIR, has secured the top position on the CyberGym benchmark, an independent evaluation for AI-driven security tools. The engine achieved a 97% success rate, surpassing the previous leader's score of 93.2% recorded on August 8, 2026.

Developed by the University of California, Berkeley, the CyberGym benchmark tests AI security tooling against 1,507 confirmed vulnerabilities sourced from 188 large open-source software projects. TrendAI utilizes seven AI models from four different providers: Anthropic, DeepSeek, Google, and OpenAI.

According to Rachel Jin, Chief Platform and Business Officer and Head of TrendAI, the goal of the technology is to compress the exposure window between vulnerability discovery and exploitation. The system aims to provide rapid risk reduction by translating validated intelligence into detection and virtual patches, bypassing traditional patching cycles.

Entities

AESIR · CyberGym · Rachel Jin · TrendAI · University of California, Berkeley

Claims

What the coverage asserts, and how many sources carry each claim.