Report Reveals Gap Between AI Use and AI Security In Embedded Software

"The State of Embedded Software Quality and Safety 2025 from Black Duck reveals a disconnect between the organizational use of AI and AI security. The embedded software landscape is transforming, largely driven by AI, with 89.3% of organizations already utilizing AI coding assistants and 96.1% integrating products with open source AI models. However, 21.1% of organizations still lack confidence in their capabilities to prevent AI from opening the door to vulnerabilities."

"Furthermore, the prevalence of shadow AI presents another risk to companies, impacting 18%. Key findings from the report include: Software Bills of Materials (SBOMs) are becoming a commercial necessity, with 70.8% of organizations producing them primarily due to customer and partner demands (39.4%). Embedded developers are finding their roles adjusted, with 80.4% of organizations adopting memory-safe languages. 86% of CTOs and directors view their projects as successful, but only 56% of hands-on developers agree."

Embedded software use of AI is pervasive, with 89.3% of organizations using AI coding assistants and 96.1% integrating products with open source AI models. Confidence gaps exist: 21.1% of organizations lack confidence in preventing AI-enabled vulnerabilities, and 18% face shadow AI impacts. Software Bills of Materials (SBOMs) are widely produced (70.8%), often driven by customer and partner demand (39.4%). Memory-safe languages are widely adopted (80.4%), shifting developer roles. Perception of project success diverges: 86% of CTOs and directors rate projects successful versus 56% of hands-on developers.

Read at Securitymagazine