Germany’s VDE testing organization announced on May 20, 2026 the launch of the ‘AI-enabled Sensor Functional Safety’ (AI-SaFe) voluntary evaluation pilot project, based on the newly released ISO/IEC 23053:2026 standard. This move has a direct transmission effect on niche sectors that rely on functional safety certification, such as automotive electronics, industrial robots, and smart sensor ODM manufacturing, marking that the compliance pathway for AI-embedded sensors in high-reliability scenarios has begun to take shape.

Event Overview

Germany’s VDE testing organization officially announced on May 20, 2026 the launch of the ‘AI-enabled Sensor Functional Safety’ (AI-SaFe) voluntary evaluation pilot project. This project is strictly based on the new international standard ISO/IEC 23053:2026 released in 2026, focusing on the functional safety assessment of core modules such as edge AI inference, self-diagnostic algorithms, and anomalous data filtering. The pilot is open globally and explicitly welcomes Chinese sensor ODM manufacturers to participate in pre-certification; companies that pass the pre-certification will receive the VDE Green Label and an official recommendation letter. The deadline for the first round of applications is June 15, 2026.

Which Segmented Industries Will Be Affected

——Automotive electronics supply chain companies
German automakers and Tier 1 suppliers generally regard VDE certification as one of the prerequisite entry conditions. If this pilot forms a reusable functional safety assessment framework for AI sensors, it will directly affect the type approval cycle and technical documentation requirements for modules such as in-vehicle millimeter-wave radar, smart cameras, and battery BMS sensors. The impact is mainly reflected in: the certification pathway extending from traditional hardware safety to AI algorithm behavior verification, with additional test dimensions such as real-time performance, robustness, and response to data drift.

——Industrial robot body and core component manufacturers
Industrial robots have clear functional safety level requirements (such as SIL2/PLd) for sensors related to force sensing, vision, and position feedback. When sensors integrate AI self-diagnosis or dynamic filtering functions, the original safety assessment model needs to adapt to new factors such as algorithm update mechanisms and tolerance for weight drift. The impact is mainly reflected in: OEMs may incorporate AI-SaFe pre-certification results into annexes of technical agreements for second-tier suppliers, making them one of the preconditions before delivery.

——Chinese sensor ODM/IDM manufacturing enterprises
The pilot explicitly allows Chinese ODM manufacturers to apply for pre-certification, making it one of the few new functional safety mechanisms in which a European authoritative institution has proactively engaged with Chinese manufacturing entities. The impact is mainly reflected in: after obtaining the VDE Green Label and recommendation letter, the technical due diligence cycle for entering the supply chains of German customers can be shortened; however, companies must simultaneously undertake new compliance documentation preparation obligations, such as algorithm explainability statements, training dataset boundary declarations, and proof of edge inference resource usage.

What Key Points Should Relevant Companies or Practitioners Focus On, and How Should They Respond at Present

Pay attention to the scope definition of the first batch of pilot results

At present, it is only explicitly stated that the evaluation will be conducted based on the ISO/IEC 23053:2026 standard, but there is still no public explanation as to whether this standard will be recognized by related regulatory parties such as the German Federal Motor Transport Authority (KBA) and TÜV Rheinland, or whether it will be incorporated into the collaborative framework of UN-R155/ISO 21448 (SOTIF). Companies need to continuously follow subsequent announcements on the VDE official website and in the EU NANDO database.

Focus on whether the AI modules already deployed in your products fall within the pilot’s coverage scope

The pilot clearly covers three types of functions: edge AI inference, self-diagnostic algorithms, and anomalous data filtering. Companies should quickly review whether any of the above modules exist in the firmware of their mass-produced sensors, with particular attention to: lightweight AI logic that is non-cloud-based, non-training-state, and runs on MCUs or low-power AI accelerators falls within the scope of this evaluation; pure software SDKs or development toolchains are not included.

Distinguish between policy signals and the actual pace of business implementation

This is a voluntary pilot, not a mandatory certification. The VDE Green Label currently does not have regulatory force, and its value is concentrated in supply chain trust endorsement. Companies should not immediately adjust their entire product roadmap, but should give priority to selecting 1–2 sensor models that have already entered the sample review stage of German customers to evaluate the input-output ratio of pre-certification.

Prepare algorithm documentation and validation record archiving systems in advance

Pre-certification requires the submission of materials such as AI module architecture diagrams, input-output data flow definitions, the basis for setting abnormal trigger thresholds, and sample response logs under at least 3 typical edge failure scenarios. It is recommended that ODM enterprises establish internal AI functional safety documentation templates with reference to Annex C of ISO/IEC 23053, and complete the traceability compilation of algorithm change records for firmware versions on the production line over the past 6 months.

Editor’s Viewpoint / Industry Observation

显然，这一试点并不是认证强制要求，而是一个制度性信号，表明嵌入式AI的功能安全治理正从理论框架转向可执行的测试协议。它反映出欧洲技术监管机构日益形成的共识：安全相关系统中的AI组件，不仅必须评估其正确性，还必须评估其在分布漂移和资源约束条件下的行为一致性。分析表明，这一时间点与欧盟《AI法案》高风险系统执法时间线（2026–2027）相吻合，暗示AI治理与传统安全标准之间正在趋于融合。然而，目前尚不清楚AI-SaFe将发展为独立的VDE认证，还是会被纳入IEC 61508-3或ISO 26262-8等现有标准的扩展版本。该试点的真正意义，与其说在于立刻带来合规压力，不如说在于揭示安全关键型AI验证正在多快地从实验室研究走向供应链准入把关。

Conclusion:
This pilot is not a mandatory market-entry threshold, but rather a mechanistic trial exploration of the European functional safety governance system extending toward AI-embedded applications. At present, it is more appropriate to understand it as: an early-stage compliance pathfinding tool for specific technical modules (edge AI sensors), whose value depends on whether it is subsequently referenced in downstream OEM procurement clauses and whether it is adopted and mutually recognized by other certification bodies. Industry participants should maintain information tracking, prudently assess the compatibility of their own product technology stacks, avoid premature comprehensive investment, and also should not overlook its guiding significance for medium- to long-term technical compliance pathways.

Information source notes:
Main source: AI-SaFe pilot announcement released by the official website of Germany’s VDE testing organization on May 20, 2026.
Areas requiring continued observation: the actual recognition status of the ISO/IEC 23053:2026 standard in regulatory practice across EU member states; and how the VDE Green Label interfaces with existing automotive/industrial safety certification systems (such as ISO 26262 and IEC 62061).