Google's AI-Powered OSS-Fuzz Uncovers Critical OpenSSL Flaw, Enhances Vulnerability Detection with LLMs
November 20, 2024Google's OSS-Fuzz project has made significant strides in vulnerability detection, identifying 26 vulnerabilities, including a critical flaw in OpenSSL that was reported in mid-September and subsequently fixed in October 2024.
Since its inception in 2016, OSS-Fuzz has continuously tested various open-source projects, and recent enhancements have introduced AI-based fuzzing to improve the coverage of codebases.
Improvements in AI capabilities have allowed for better context generation and simulation of developer workflows, which contributed to the successful detection of these vulnerabilities.
The automated fuzzing process has become more accurate and efficient, leading to a greater number of correct fuzz targets being identified.
Google's AI-driven fuzzing tool works by injecting random data into software, effectively uncovering errors that may be overlooked by human testers.
Google aims to fully automate the fuzzing workflow, including the generation of patches for discovered vulnerabilities, and is collaborating with researchers to advance this technology.
Over the past 18 months, enhancements to OSS-Fuzz have resulted in increased code coverage for 272 C/C++ projects, adding over 370,000 new lines of code.
Earlier in November 2024, Google announced that its LLM-based framework, Big Sleep, detected a zero-day vulnerability in the SQLite database engine, which had been undetectable by traditional methods.
The integration of large language models (LLMs) has further automated fuzzing workflows, enhancing the effectiveness of vulnerability detection.
In addition to fuzzing improvements, Google is transitioning its codebases to memory-safe programming languages like Rust and retrofitting existing C++ projects to mitigate spatial memory safety vulnerabilities.
Recognized for its advanced security research, Google frequently identifies vulnerabilities in its own products, such as Chrome and Gmail, showcasing its commitment to software security.
In response to the critical OpenSSL vulnerability, updates have been released to patch the issue, demonstrating the ongoing efforts to secure software against emerging threats.
Summary based on 5 sources
Get a daily email with more AI stories
Sources
Forbes • Nov 20, 2024
Google Confirms Critical 20-Year-Old Security Flaw Using New Fuzzy AITechRadar pro • Nov 21, 2024
Google's AI-powered bug hunting tool finds a host of concerning open source security flawsThe Hacker News • Nov 21, 2024
Google's AI-Powered OSS-Fuzz Tool Finds 26 Vulnerabilities in Open-Source ProjectsThe Register • Nov 20, 2024
Google's AI bug hunters sniff out two dozen-plus code gremlins that humans missed