AI, ready to deploy and be trusted

UL 3115 is the OOI for the Safety of AI‑Based Products. It provides a structured approach for evaluating AI safety across a wide range of products and systems, focusing on how AI is designed, developed and deployed rather than on specific algorithms or use cases.

Through AI safety certification aligned with UL 3115, UL Solutions conducts an independent, third‑party assessment to evaluate whether an AI‑based product has implemented appropriate safety measures based on its intended use and risk profile. The assessment focuses on requirements and evidence, while responsibility for design and operation remains with the organization.

UL 3115 and the NIST AI RMF address AI safety from complementary perspectives. The NIST framework provides voluntary, organization‑level guidance for identifying and managing AI risks across the life cycle. UL 3115 applies those concepts at the product level, offering a structured framework for evaluating whether a specific AI‑based product meets defined safety expectations for its intended use. Used together, UL 3115 and NIST AI RMF can help organizations manage AI risk consistently and demonstrate product‑level safety in practice.

ISO/IEC 42001 is an organization‑level standard for AI management systems. It sets requirements for governance, policies, roles and processes that guide how an organization develops, deploys and manages AI across its activities.   
 
UL 3115, by contrast, specifies safety requirements and provides guidelines for safer design, development and deployment of AI-based products. It covers the following governance-related AI safety principles: data collection and data analysis, documentation and reporting, life cycle management, accountability, control and oversight, and transparency and explainability. These principles are applied while evaluating whether a specific AI‑based product is safe for its intended use. 

AI safety certification evaluates an AI‑based product’s compliance with defined safety expectations based on its intended use and risk profile, while MCV focuses on verifying the accuracy of specific AI‑related marketing claims for algorithm reproducibility, data privacy and model transparency through an objective, science‑based assessment. While both draw on shared AI safety principles, they serve different purposes. 

Operationalizing AI life cycle governance involves establishing clear processes for oversight, documentation and monitoring from design through deployment and beyond. This includes defining roles, managing change and maintaining records that support accountability as AI systems evolve over time.

Yes, UL Solutions can evaluate a wide range of AI-based products and systems, including those operating in safety‑critical contexts. The applicable approach depends on the intended use, risk profile and regulatory expectations associated with the specific application.

AI stewardship and marketing claim verification help organizations demonstrate responsible AI practices beyond compliance. Third-party verification and clear communication about how AI systems behave and what claims are being made can help reduce uncertainty and strengthen confidence among customers, partners and regulators.