Google Project Zero and Google DeepMind report that Big Sleep, an LLM agent evolved from Project Naptime, found an exploitable stack buffer underflow in SQLite during a variant-analysis experiment seeded with newly landed commits. The bug was reported in early October 2024 and fixed the same day, before it appeared in an official release; Google believes this was the first public example of an AI agent finding such a memory-safety flaw in widely used software.
Why it matters
It marked the move from benchmark results to a real, previously unknown vulnerability found by an LLM agent and fixed before users were exposed.
Key facts
As stated in the sources, with where to find them.
- The vulnerability was found and reported in early October and fixed by SQLite developers the same day; it never reached an official release.Introduction
- Existing testing infrastructure, including OSS-Fuzz and SQLite's own, had not found the issue.Introduction
- The agent was given newly landed SQLite commits and diffs and asked to look for related unfixed issues at HEAD (variant analysis).Section 'Methodology'
Findings that cite this record
Key questions this bears on
- Is AI shifting the balance between finding and fixing vulnerabilities?Discovery is ahead. Verification, triage, coordinated disclosure, safe repair and getting fixes to users remain bottlenecks between AI findings and protection.
- Can AI agents defend and oversee systems on their own?Not reliably. Some designs improve benchmark triage, but autonomous defense lacks operational validation and production oversight of agents has documented gaps.
Sources
Related records
Jul 15, 2025
Aug 4, 2025
Oct 27, 2026