Webinar Description
Key Takeaways
- Security pass rates for AI-generated code have remained at 56% across more than 100 models tested since 2023, despite improvements in code syntax.
- The webinar presents findings from the 2026 GenAI Code Security Report, drawing on nearly four years of data.
- Designed for security leaders, developers and risk managers in enterprises scaling AI-assisted development.
- Focuses on verification, release controls and accountability rather than developer tooling.
- Covers vulnerability classes, Java performance and dependency risk specific to AI-authored code.
About the Webinar
Veracode hosts this live webinar to examine the security implications of AI-generated code, using data collected from 2023 to 2026 and published in the 2026 GenAI Code Security Report. The session addresses a central finding: while AI models have improved at producing syntactically correct code, security outcomes have not kept pace. The 56% security pass rate observed across more than 100 models represents a persistent challenge for organisations adopting AI-assisted software development at scale.
The webinar positions the problem as one of verification and governance rather than model selection. Presenters will explain why choosing the “best model” does not resolve underlying security gaps and why verification processes must operate independently of the AI systems generating the code.
Vulnerability Classes and Risk Models
Attendees will learn how to interpret differences across vulnerability classes in AI-generated code. The session covers how these patterns affect risk prioritisation and remediation strategies. Java performance and dependency risk receive specific attention, reflecting common enterprise technology stacks where AI code generation is increasingly deployed.
The webinar also addresses how stagnant security pass rates change the risk model for organisations scaling AI-assisted development, requiring updated approaches to policy, accountability and release controls.
Verification and Release Controls
A core theme of the session is that verification must exist outside the AI model itself. The webinar explores practical steps for improving risk visibility across the software delivery pipeline, including how to implement release controls that account for AI-authored code. Testing methodologies discussed include SAST, DAST, SCA and container security, alongside AI code remediation techniques.
Who Should Attend
The webinar targets CISOs, heads of application security, developers, security teams and IT risk managers. It is relevant to organisations in software, financial services, government, healthcare, retail and energy sectors that are adopting or expanding AI-assisted development. The content suits both executive and technical audiences seeking to translate research findings into actionable security improvements.