The EU AI Act remains the subject of many controversies. While some see it as an obstacle to innovation, others still celebrate it as a lighthouse victory for the responsible use of artificial intelligence. Regardless of what side of the fence you sit on, 2025 will be a big year for the EU AI Act! And at QuantPi we are ready to support organizations in preparing for compliance, by offering technical assessments to test their AI models against the regulatory requirements.
With fast approaching deadlines, we are reflecting on a panel discussion we hosted last year with experts: Vera Sikes from the German Federal Office for Information Security, Isabel Gadea from Deloitte and Felix Broßmann from SKAD AG. They shed light on the areas of application, opportunities and challenges of the EU AI Act. As well as, derived recommendations on how companies can implement the Act, whereby interdisciplinary cooperation between AI developers and compliance teams is crucial.
Read on for a snapshot of the topics covered and challenges to consider for organizations looking to navigate AI compliance successfully.
Scope of the EU AI Act: It is not only the major providers of AI models that must comply with the provisions of the EU AI Act. Companies using general purpose models for their products and making changes to them can also be affected by the EU AI Act.
Risks and Liability: The introduction of the EU AI Act may bring legal challenges similar to the GDPR, particularly regarding class action lawsuits and liability issues, which forces companies to take proactive measures.
So long AI “Wild West”: Regulations are shaping up all over the world. Being proactive in regards to setting up responsible AI governance practices and processes will benefit organizations immensely down the line. The regulatory landscape in the USA, for example, though increasingly fragmented, already has a large number of state laws which apply to AI.
Balancing innovation and compliance: Successful implementation depends on interdisciplinary collaboration between departments such as compliance, legal, and AI development. Efficient collaboration between these areas is critical.
Testing blackboxes: Regulatory standards must be translated into test criteria that take into account the blackbox nature of AI models. In addition, a suitable measurement technology is necessary to test the model for these criteria to ensure trustworthiness.
Making the most out of limited resources: Organizations need to actively engage with the topic of AI governance and bundle their often limited resources of AI expertise as efficiently as possible. From an organizational perspective, assigning clear responsibilities for AI Governance is essential to ensure that the rapid technological and regulatory changes in the AI landscape can be implemented quickly and safely.
Should you want to watch the on-demand recording, please note the panel discussion is in German.
QuantPi pioneers trustworthy AI with its holistic platform for rigorously testing AI systems for bias, robustness, compliance and other critical performance metrics, ensuring transparency and actionable insights for successful AI transformations. Backed by the EU and emerging from CISPA, QuantPi advances responsible AI globally, building confidence in intelligent systems.
For more information, email contact@quantpi.com