On 28 May, Plattform Industrie 4.0 hosted the seventh edition of the webinar AI Use Cases & Future Skills, organised as part of the Interreg AT-SK project Twin City Future Innovation Manufacturing Hub under the thematic focus of Work 5.0.
Speakers Rania Wazir (leiwand.ai) and Thomas Doms (TÜV AUSTRIA) provided key insights into how artificial intelligence can be designed to be safe, explainable, and trustworthy.
AI standards as a foundation for the AI Act
In her presentation, Rania Wazir, Co-Founder and CTO of leiwand.ai, explained the central role of standardisation in implementing the European AI Act. The regulation aims to ensure a high level of protection for health, safety, fundamental rights, and democracy. Standards are intended to help companies translate these requirements into practice.
The focus lies on the so-called “essential requirements” of the AI Act, including risk management, data quality, transparency, human oversight, robustness, accuracy, and cybersecurity. Harmonised European standards are currently being developed for these areas. It became clear that standardisation is a complex and time-consuming process. Although the European Commission mandated the development of these standards in 2023, many documents are still under consultation or in coordination phases.
The first full set of standards is expected to become available from 2027 onwards. A key challenge is translating the legal requirements of the AI Act into technical quality characteristics. Using “accuracy” as an example, it was shown that raw model accuracy is often insufficient. More important is whether a system reliably performs its intended task while treating different user groups fairly. Standards help define appropriate metrics, make test results comparable, and ensure transparency, thereby providing a foundation for developing, assessing, and certifying trustworthy AI systems.
AI at TÜV AUSTRIA
In his presentation, Thomas Doms, Head of Research, Development & Innovation at TÜV AUSTRIA, demonstrated how the organisation not only evaluates and audits AI systems but also actively uses AI in its own operations. Through its subsidiary Trustify, TÜV AUSTRIA—together with the Software Competence Center Hagenberg (SCCH)—supports companies in building safe, compliant, and trustworthy AI systems. This includes audits, compliance checks, and assessments of data protection, security, and quality requirements.
AI also plays an important role within the organisation itself. In addition to standard solutions, TÜV AUSTRIA develops its own AI applications to improve efficiency and support employees. Current use cases include an AI-powered expert assistant for generating inspection reports, AI support in the Cyber Security Operations Center, and the new website assistant “Adam”, which helps users find information and services.
The development and evaluation of these systems is guided by an internal AI assessment catalogue comprising around 400 criteria. Key focus areas include data quality, functional correctness, cybersecurity, transparency, and human oversight. A core principle remains unchanged: in safety-critical decisions, responsibility stays with humans. AI supports experts but does not replace them.
The presentation emphasised that trustworthy AI requires not only advanced technology, but also clear processes, quality standards, and responsible implementation.
