Google arregló más bugs de Chrome en junio que en los dos años anteriores juntos, gracias a la IA

Summary

In June 2026, Google closed 1,072 Chrome security vulnerabilities using a Gemini-based artificial intelligence system—more than the total from the previous two years—and this system detects, validates, and suggests patches for these vulnerabilities.

Why it matters

This demonstrates a dramatic acceleration in vulnerability discovery, highlighted by the detection of a sandbox escape that had persisted for 13 years. AI-assisted workflows have the potential to scale and reshape the cybersecurity economy by reducing false positives and developer workload.

Highlights

• AI agents performed semantic code analysis and generated multiple patches for each validated vulnerability.
• The system automatically prioritized the surge in external bug bounty reports, filtering spam and assigning severity levels.
• Google reported that AI identified a long-standing sandbox escape capable of compromising an entire operating system.

Critical take

While AI assistance speeds up detection, it also narrows the security gap for attackers who can wield the same tools; this raises the possibility that the 1,072 fixed bugs could simultaneously become an exploitable set of vulnerabilities.