TrendAI says AESIR ranks first on CyberGym AI security benchmark
TrendAI said its AESIR exploit-remediation engine ranked first on the CyberGym benchmark with a 97% success rate.
TrendAI said on Aug. 26 that its AESIR agentic exploit-remediation engine ranked first on the CyberGym benchmark, a University of California, Berkeley evaluation for AI-driven security tooling.
The company said AESIR posted a 97% success rate, nearly 4 percentage points higher than the former leader’s 93.2% score reported on Aug. 8. TrendAI said the result shows AI systems can do more than find vulnerabilities and can also help validate and reduce risk faster.
According to the release, CyberGym tests AI security tools against 1,507 confirmed vulnerabilities from 188 large open-source software projects. TrendAI said its system uses seven AI models across four providers: Anthropic, DeepSeek, Google and OpenAI.
The company said AESIR identifies weak spots, validates findings, turns them into working proof of concept, and then reduces risk with detection and virtual patches without waiting for the normal patching cycle. TrendAI said the capability is built on more than 20 years of vulnerability research and exploit intelligence from TrendAI ZDI.
Rachel Jin, identified in the release as chief platform and business officer and head of TrendAI, said CyberGym’s validation changes what is possible for enterprises adopting AI. Jin also said the goal is to compress exposure time toward zero.
TrendAI said the engine also supports dynamic discovery of threat actors actively exploiting vulnerabilities in the wild. The company said it published a technical blog post on the CyberGym methodology.
TrendAI said the benchmark result is current as of the Aug. 26 announcement.
Sources
Editorial note: This article was prepared with AI assistance and includes source attribution for reader review.
View the Web StoryA visual, tap-through version of this reportThis is the AMP version of the article. Read the full article on Atlas Watchline.