Report by Sonatype

The AI-Era Software Assembly Line

10 FINDINGSPublished Aug 18, 2026
View Original Report →

Key Findings

Newly affected software component versions increased at 46 times the pre-AI rate.

Dependency ManagementVulnerabilitiesOpen Source

Critical and High-severity vulnerabilities per enterprise application increased 4.31x.

VulnerabilitiesEnterprise ApplicationsSoftware Security

The median age of unresolved Critical and High vulnerabilities fell to 103 days by May 2026.

VulnerabilitiesRemediation

For vulnerable dependencies introduced into AI-era applications, a materially lower-risk PyPI version was already available 34.3% of the time when the dependency was selected.

Dependency ManagementPyPIOpen SourceVulnerabilities

For vulnerable dependencies introduced into AI-era applications, a materially lower-risk npm version was already available 46.9% of the time when the dependency was selected.

Dependency ManagementnpmOpen SourceVulnerabilities

Average Critical and High-severity vulnerabilities per analyzed application increased from 14.14 in June 2022 to 54.3 by 2026.

Vulnerabilities

Median vulnerability age fell 59% from its January 2024 peak.

VulnerabilitiesRemediation

The median age of unresolved Critical and High vulnerabilities decreased by 45%, dropping from 228 days to 126 days during the AI era.

VulnerabilitiesRemediationTime-to-Fix

Among early AI-era Critical and High vulnerability cohorts with at least 12 months to remediate, 52.6% were resolved, 44.3% remained open, and 3.1% were waived.

VulnerabilitiesRemediation

Even after excluding newly managed legacy applications, Critical and High vulnerability risk per application increased 3.91x.

VulnerabilitiesLegacy Systems