Following earlier work on ENACT’s ethics-by-design approach and the co-creation of use cases, the project is now moving into a new phase: validating how ethical and legal requirements hold up in practice.
As part of ongoing work under ENACT’s ethics and privacy framework, partners from LAUREA are advancing the Z-inspection® process. This step plays a key role in ensuring that the project’s AI systems are not only designed responsibly but also critically assessed against real-world assumptions and constraints.
The Z-inspection® process is an internationally recognized method for assessing the trustworthiness of AI systems. Within ENACT, it supports our ethical work by helping identify both ethical and legal risks and any gaps in existing requirements. It also complements the “ethics-by-design” approach, which encourages project teams to reflect early on how the technology fits into society and how different users and stakeholders interact with it.
Testing assumptions through sociotechnical scenarios
At the core of this process, each use case team develops sociotechnical scenarios. These describe, in simple terms, which AI and digital tools are involved, who uses the system, and how the information produced by the system flows between different actors.
An expert group, bringing together specialists from fields such as ethics, technology, and clinical practice, then reviews these scenarios. Their task is to identify claims or assumptions that need to be tested. The use case teams respond by providing arguments and evidence to support these claims. If evidence cannot be found, the topic may be flagged as a potential risk that requires further assessment.
This process builds directly on earlier project work, such as technical requirements and ethical or legal desktop studies. The Z-inspection® process adds value by highlighting any issues that were not previously captured, for example, a claim that the system performs actions that are not technically possible, or a need for a new technical requirement that must later be validated.
Strengthening project outcomes across ENACT
The Z-inspection® process plays an important role in connecting different strands of work across ENACT. By feeding its findings back into requirement setting and system design, it helps strengthen the development of AI models and supports more robust integration into the project’s platform.
The reviews for the studies in Belgium, Bulgaria, and Ireland are expected to be completed by the end of May. Their findings will directly contribute to improving how ethical and legal risks are identified, assessed, and mitigated across the project.
Ultimately, this work ensures that ENACT’s approach to AI does not stop at defining requirements but actively validates them, supporting the development of trustworthy solutions that are aligned with both technical realities and societal needs.