Safety of AI-Based Tools: Theory and Practical Experiences

AI is creating new opportunities to improve software development efficiency, but organizations developing safety-related systems must ensure AI-enabled tools are used safely, compliantly, and with appropriate safeguards. As AI-generated code recommendations become increasingly common, understanding how to validate and trust these outputs is critical.

Join experts from Validas AG and Perforce Software for a discussion on how AI-based tools can be deployed safely in regulated development environments, and the role static analysis and formal methods play in safeguarding AI usage in safety-related systems. The session will also demonstrate how AI-powered refactoring recommendations can be generated from static analysis findings and applied with greater confidence.

You'll also gain insight into emerging AI safety standards, qualification requirements, and verification approaches that can help organizations adopt AI while maintaining the safety and compliance expectations required for safety-critical software.

In this webinar, Dr. Oscar Slotosch, CEO of Validas AG, and Steve Howard, Director of Product at Perforce Software, will discuss how development teams can:

• Apply tool classification and qualification principles from functional safety standards, including emerging requirements from ISO/IEC TS 22440 for AI-enabled development tools.

• Use Perforce Static Analysis as the foundation for AI-assisted remediation, including a demonstration of how a defect discovered in the OpenSSL project can be automatically transformed into an AI-generated refactoring recommendation.

• Assess the risks of AI-generated code changes and leverage formal methods, including the Validas R3 prototype, to automatically verify that AI-proposed fixes resolve detected violations without introducing unintended side effects.

Register now to learn how static analysis, AI-assisted remediation, and formal verification can help organizations safely adopt AI while maintaining confidence in the quality, behavior, and compliance of safety-critical software.