ISO/IEC TR 5469:2024

ISO/IEC TR 5469:2024 addresses a critical challenge in modern technology: how to safely integrate artificial intelligence into systems where failure could cause harm to people, property, or the environment. Published in January 2024, this technical report bridges the gap between traditional functional safety engineering and the emerging complexities of AI deployment.

​Integration of AI and Functional Safety

​Understanding the Scope: Three Core Applications

The technical report explicitly covers three distinct scenarios for AI and functional safety integration: using AI components to directly implement safety-critical functionality, employing traditional safety mechanisms to ensure AI-controlled systems remain safe, and leveraging AI tools to design and develop traditional safety-related functions. This tri-fold approach recognizes that AI's role in safety systems isn't monolithic, meaning different applications require tailored risk management strategies.

​

Technical Report vs. Standard: What This Means

ISO/IEC TR 5469:2024 is designated as a Technical Report (TR), not a normative standard. This classification is significant because it provides informative guidance and best practices rather than mandatory requirements. It captures current industry learning as the field evolves and serves as a foundation for future standards, such as the forthcoming ISO/IEC TS 22440. Organizations can use this report to inform their AI safety strategies, though compliance is not mandatory or certifiable under this document alone.

​

Key Technical Components

The report structures AI safety considerations around a three-stage realization principle. The Data Acquisition Stage focuses on input validation and sensor reliability, while the Knowledge Induction Stage involves training data verification and model validation techniques. Finally, the Processing and Output Stage addresses inference reliability and safety constraints. This framework parallels the traditional sensor-controller-actuator model, making it accessible to safety engineers familiar with standards like IEC 61508 or ISO 26262.

The report outlines three primary risk mitigation approaches for AI embedded in products. Primary approaches include Backup Systems, which implement non-AI fallback functions for when AI components fail, and Supervisory Controls that create boundary systems to constrain AI outputs. Additionally, Redundant AI Voting can be deployed, using multiple AI models to identify and reject anomalous outputs to maintain system integrity.

Integration with AI Lifecycle Management

A critical aspect of the report is its emphasis on aligning functional safety lifecycles with AI lifecycle processes. The document references ISO/IEC 5338 for AI system lifecycle processes, IEC 61508 for functional safety lifecycle alignment, and ISO/IEC 42001:2023 for AI management system requirements, emphasizing patents and conformity assessment types. This integrated approach ensures safety considerations are embedded throughout AI development, deployment, and maintenance phases.

Verification and Validation Framework

​The report also provides a robust framework for Verification and Validation (V&V) activities specifically tailored to AI-safety integration. Verification methods involve statistical performance validation, robustness testing against adversarial inputs, and explainability assessments for safety-critical decisions. These ensure the model's behavior is as transparent and resilient as possible before deployment.

To manage the system during operation, the report outlines control measures such as runtime monitoring systems, graceful degradation protocols, and strict human oversight requirements. These are supported by process methodologies that include safety case development for AI components and risk assessment adaptations for non-deterministic systems. Comprehensive documentation requirements for AI training and validation ensure that the entire safety argument is traceable and verifiable.

ISO/IEC TR 5469:2024 Framework: From AI Safety Principles to Implementation

Industry Applications and Limitations

While ISO/IEC TR 5469:2024 is industry-agnostic, its practical application varies significantly across different sectors. It is particularly well-suited for automotive advanced driver assistance systems and industrial automation using AI-enhanced controls. It also provides a valuable framework for medical devices incorporating AI diagnostics and robotics that utilize machine learning capabilities. By offering a cross-industry foundation, it allows these various fields to apply a consistent logic to AI safety.

However, the report has notable current limitations. It lacks specific quantitative safety integrity level (SIL) mappings for AI, which are often required for high-stakes certification. It also does not address formal certification pathways for AI safety or provide extensive guidance on continuous learning systems, where models evolve after deployment. These gaps highlight the document's role as an informative guide rather than a final regulatory requirement.

​For organizations looking to implement these guidelines, the first step is to assess current safety processes to evaluate how existing functional safety procedures can accommodate AI uncertainty. This should be followed by developing AI safety competencies, specifically training safety engineers in AI-related risks and concepts. Organizations must also prioritize documenting AI design decisions, creating traceable records that link AI choices directly to safety requirements. Finally, establishing monitoring systems is essential to implement continuous performance tracking for AI safety functions, ensuring the system remains within safe operating parameters over time.

​Taking Action: Next Steps for Safety Professionals

ISO/IEC TR 5469:2024 provides essential guidance for organizations navigating AI integration in safety-critical applications. While not mandating specific requirements, it establishes a framework for systematic risk assessment and mitigation. Safety professionals should view this technical report as a roadmap for responsible AI deployment in safety systems. Begin by mapping your current AI applications against the three scenarios outlined, identifying gaps in your safety assessment processes, and implementing the verification and validation practices detailed in the report.

The fundamental goal of protecting human life and well-being. By providing gap analysis and specialized training, Nemko Digital ensures safety is maintained throughout the AI lifecycle. Contact us and discover how our AI Trust Services can secure your path to compliance.