Microsoft’s AI Uncovers 99 Security Issues in Express, ImageMagick and Other Open-Source Projects

Blog WriterCybersecurity News - Original News Source is cybersecuritynews.com


Microsoft and GitHub sent 99 verified security reports to six open-source projects, including Express and ImageMagick, after an AI pilot identified thousands of potential vulnerabilities.

In its Nov. 17 announcement, GitHub said maintainers had fixed 55 issues as of Nov. 15. However, only 14 of those fixes had made it into software releases.

The other four projects, Caddy, AutoGPT, FastMCP, and Netty, also participated in the GitHub Secure Open Source Fund, with maintainers approving the scans before the pilot began.

Microsoft’s Codename MDASH scanned the projects with an earlier version of the MAI Cyber model and suggested code fixes. A GitHub Security Lab AI tool removed duplicates, filtered out reports that didn’t hold up, and prioritized higher-impact issues.

Seven Microsoft FORGE Lab researchers then manually reviewed the remaining findings. Microsoft had previously used MDASH to find Windows vulnerabilities.

Codename MDASH AI Flags Thousands of Potential Flaws (source : github )
Codename MDASH AI Flags Thousands of Potential Flaws (source : Github )

Microsofts AI Uncovers 99 Security Issues

GitHub sent the verified reports through its private vulnerability reporting system. Each report included evidence of the issue and a proposed patch.

According to GitHub, maintainers could accept, modify, or replace suggested patches and decide when to release fixes without receiving raw scan results. Caddy fixed all 22 reported issues within a week, while ImageMagick’s two-person team also received 22 reports.

One looked like a problem the team had seen before, but the report explained what was different and pointed to the relevant code. The team pushed a fix in 15 minutes. Those findings are separate from previously reported ImageMagick memory corruption flaws.


maintainers received 99 private reports (source : github )
Maintainers received 99 private reports (source: GitHub)

AutoGPT’s team received a report about a newly added security check three days after it shipped. The maintainers went beyond the suggested patch, wrote their own fix within a day, and put it into production in less than a week. Express maintainers said the reports helped them cut through the noise from other vulnerability submissions.

The pilot covered only six projects, and the human review took about two days per project. That makes the results an early test of how AI can help small teams, not proof that the approach can work at the same pace across thousands of repositories.

The work also does not end when a maintainer writes a patch. Users need a release that contains the fix, and they still need to install it.

GitHub plans to offer reviewed AI scans to all alums of its Secure Open Source Fund and deliver findings privately. For developers who rely on open-source packages, GitHub recommends enabling Dependabot alerts to catch known vulnerabilities in their dependencies.

Once a public advisory is reviewed for the GitHub Advisory Database, Dependabot can notify repositories that use the affected package. Teams can then check the advisory against the version they actually run and confirm that their next update includes the fix.

Give your SOC the alert context on the latest threats to stop incidents quickly -> Integrate Threat Intel from 16K SOCs

The post Microsoft’s AI Uncovers 99 Security Issues in Express, ImageMagick and Other Open-Source Projects appeared first on Cyber Security News.