Google AI “Big Sleep” Identifies Critical SQLite Vulnerability Before Hackers Can Exploit It
July 16, 2025
AI Security / Vulnerability
Google announced on Tuesday that its language model-assisted vulnerability detection system successfully identified a security flaw in the SQLite open-source database engine, preventing potential exploitation. The vulnerability, designated CVE-2025-6965 (CVSS score: 7.2), is a memory corruption issue affecting all versions prior to 3.50.2. Discovered by “Big Sleep,” an AI agent developed through a collaboration between DeepMind and Google Project Zero, this flaw allows for potential attacks through arbitrary SQL statements, leading to integer overflow risks. SQLite maintainers cautioned that this critical security issue was previously known only to threat actors. Google has not disclosed the identities of these actors but emphasized the urgency of addressing the vulnerability.
AI Security / Vulnerability
Google AI “Big Sleep” Detects Critical SQLite Vulnerability Before Exploitation Could Occur On July 16, 2025, Google announced a significant achievement in cybersecurity through its AI-driven vulnerability assessment tool, known as Big Sleep. This large language model (LLM)-assisted framework successfully…