AI Safety Certification and Stewardship Services

The European Union (EU) AI Act introduces a risk‑based framework that sets out different obligations depending on how and where AI is used. High‑risk AI systems are subject to additional expectations related to risk management, documentation, human oversight and post-deployment monitoring. Even organizations operating outside the EU may be affected if AI‑enabled products or services enter the European market.

The General Data Protection Regulation (GDPR) continues to influence how organizations collect, use and protect data within AI systems. Requirements related to data protection, transparency and accountability directly apply to and shape AI development and deployment, particularly when systems rely on personal data or automated decision-making.

In the United States, the National Institute of Standards and Technology (NIST) AI Risk Management Framework (RMF) provides voluntary guidance for identifying, assessing and managing AI‑related risks. The framework emphasizes governance, technical safeguards and ongoing monitoring across the AI life cycle. While not a regulation, it is increasingly referenced as a common baseline for responsible AI practices. 

Alongside regulatory activity, international standards bodies continue to develop guidance on AI safety, governance and reliability. These efforts reflect a shared goal of supporting innovation while reducing the risk of unintended harm. Alignment with relevant regulations, frameworks and standards helps organizations apply consistent approaches to governance, risk and oversight, supporting interoperability, clearer expectations and greater trust as AI systems evolve and scale.

UL 3115 addresses AI safety through technical, ethical and governance considerations. It looks at how AI systems are designed, developed and deployed, with attention to how they behave in real‑world conditions. This includes factors such as robustness, risk management, transparency, data handling and oversight across the AI life cycle.

We conduct independent, third-party assessments aligned with UL 3115 to evaluate whether an AI-enabled product or system has implemented appropriate safety measures for its intended use and risk profile. The evaluation considers how safety is addressed in design, governance and ongoing oversight, as well as how the system behaves in real-world operation, examining outcomes alongside underlying algorithms and architecture. This approach provides objective insight into system reliability, predictability and resilience, helping organizations demonstrate that AI safety considerations are integrated throughout the product life cycle as systems evolve and scale.

AI systems evolve over time, which makes governance and documentation critical. UL 3115 considers how organizations manage change, maintain records and monitor performance throughout the AI life cycle. This includes anticipating future states, tracking risks and supporting accountability.

Independent, third-party verification of AI‑related marketing claims helps organizations communicate system performance and behavior more clearly and credibly. UL Solutions’ Marketing Claim Verification (MCV) services help manufacturers substantiate claims related to algorithm reproducibility, data privacy and model transparency, supporting clearer communication with customers, partners and regulators.

Many organizations have strong technical talent but limited internal experience with global AI safety standards, AI governance expectations and evaluation practices. AI training services can help teams build a shared understanding of risk, oversight and documentation — so they can deploy AI systems with clearer decision-making and accountability.