AI is rapidly advancing in cybersecurity, demonstrating hacking skills that experts say are nearing a critical "inflection point." RunSybil, a cybersecurity startup, discovered this in November when its AI tool, Sybil, identified a complex vulnerability in a customer's system. The AI detected a flaw in the customer's deployment of federated GraphQL, a language for accessing data via APIs.

Sybil, which uses various AI models and proprietary techniques, found the customer was unintentionally exposing confidential information. The discovery was significant because it required a deep understanding of multiple systems and their interactions. RunSybil cofounders Vlad Ionescu and Ariel Herbert-Voss confirmed that Sybil identified the issue before it was publicly known.

This incident highlights AI's increasing ability to reason and identify sophisticated security risks. RunSybil has since found the same GraphQL problem in other deployments. The company believes this marks a substantial leap in AI's cybersecurity capabilities.

RunSybil's Sybil scans computer systems for vulnerabilities that hackers could exploit. This includes unpatched servers and misconfigured databases. The company is now focusing on refining Sybil's abilities to proactively address emerging threats.

Experts predict AI will play an increasingly crucial role in both offensive and defensive cybersecurity strategies. The industry anticipates further advancements in AI-driven security tools in the coming months.