Grego AI launches with AI-powered vulnerability detection

Grego AI, a startup focused on artificial intelligence for cybersecurity, has officially launched its service. The company claims its unique approach utilizes existing AI models to identify critical software vulnerabilities that often go unnoticed by human auditors and other automated tools, as reported by Silicon Angle.

The company's method, called Deep Invariant Analysis, scans entire codebases to map module and dependency connections. It deploys small agents into sandboxed environments, each traversing different paths through the code. If an agent detects a potential weakness, it generates a proof-of-concept exploit to confirm reproducibility. Grego AI targets complex flaws arising from multiple layers of dependencies, which are typically missed by human review and traditional testing tools. The company highlighted its role in identifying a vulnerability in a major blockchain protocol, which could have led to a $27.7 million loss. This discovery resulted in a $250,000 bug bounty, reportedly the largest awarded for an AI-discovered flaw.

While initially focusing on Web3 due to its high-stakes environment, Grego AI plans to expand its services to conventional enterprise software, including financial, healthcare, cloud, and government systems.

Source: Silicon Angle