Google has revealed that its AI-based vulnerability detection tool has uncovered 20 security flaws across several popular open-source and commercial software projects.
The announcement marks a significant step forward in how artificial intelligence is being used to improve cybersecurity.
Out of the 20 identified issues, 11 have already been confirmed and fixed. The remaining vulnerabilities are currently being addressed. The bugs were found in widely used components such as FFmpeg, curl, and libexpat—tools that are often integrated into operating systems, cloud services, and media platforms.
The AI tool, which leverages large language models (LLMs), analyzes code at scale to detect subtle patterns that could signal potential security risks. This approach allows for faster and broader coverage compared to manual code reviews alone.
This initiative is part of Google’s Secure AI Framework (SAIF), which focuses on enhancing the safety and reliability of AI technologies and the systems they interact with. Google believes that integrating AI with traditional security research can significantly speed up the discovery and resolution of critical vulnerabilities.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
