CertiK Launches AI Audit Tool, Says It Detected 88.6% of Vulnerability Causes

Global Web3 security firm CertiK has launched an artificial intelligence-based security auditing tool.

CertiK said July 7 that it rolled out its AI-powered audit solution, AI Auditor, along with an open-source integrated environment for AI coding agents.

AI Auditor is designed to automate Web3 security audits from the earliest stages of development. It integrates auditing into the development workflow instead of leaving it to the final stages of a project, enabling real-time security checks.

In tests based on 35 Web3 security incidents that occurred in 2026, AI Auditor identified 88.6% of the root causes of vulnerabilities, according to the company. The evaluation used a separate validation dataset that was excluded from the model's training process.

A low false-positive rate was a core design goal. CertiK said the tool is built to address a common problem with existing security products, where excessive alerts can make real threats harder to identify, by surfacing only the most critical vulnerabilities.

AI Auditor operates on a multi-stage validator architecture. Multiple scanners detect vulnerabilities simultaneously, then pass them through a verification process that removes meaningless alerts and isolates threats with real exploitation potential.

The tool is also designed to reflect the latest security environment through continuous updates to its threat knowledge base using real-world attack cases and the latest vulnerability data.

CertiK said the tool is intended to support, not replace, security experts. It automates repetitive detection and classification tasks so specialists can focus on more complex risk analysis.

CertiK co-founder Ronghui Gu said the key question is no longer whether AI can find vulnerabilities, but how quickly it can help developers identify the risks that deserve attention. He added that AI Auditor is focused on reducing unnecessary alerts and delivering actionable security signals.