The EU AI Act prohibits AI systems that pose “unacceptable risks” to safety, health or fundamental rights. Since then, companies that develop or use AI must assess their systems according to the degree of risk and take appropriate measures to meet the legal requirements.
Article 4 of the EU AI Act also stipulates that providers and operators of AI systems must ensure that all persons involved in the development or operation of AI systems have a sufficient level of AI competence. However, the law does not provide a precise definition of “AI competence” or “AI literacy”.
In view of the many announcements from the USA and China regarding Project Stargate and Deepseek as well as Qwen in recent days and weeks, the imprecise formulation of terms in the EU AI Act has led business and science to criticize the European legislator.
The introduction of the EU AI Act and the Product Liability Directive, which has been in force since December 2024, pose considerable challenges for retailers and the FMCG industry. The Retail.ai Academy, based in Frankfurt am Main, emphasizes the need not only to understand these regulations, but also to implement them proactively and pragmatically.
While sections 4 “Literacy” and 50 “Transparency” of the EU AI Act seem sensible and the regulation of high-risk applications is generally understandable, the lesser-known new product liability guidelines are coming into focus. These require proactive and formal documentation, with the reversal of the burden of proof further exacerbating the situation.
It is alarming that successful companies such as Black Forest Labs are relocating their headquarters from the Black Forest to Delaware, USA. In addition, innovative solutions such as Meta’s Llama 4 or OpenAI’s agent solution are not finding their way to Europe. Increasing regulation written in the same spirit must be prevented as a matter of urgency.
Against this backdrop, the Retail.ai Alliance, a council of experts represented by Mark Michaelis (founder and CEO of Retail.ai GmbH), Prof. Dr. Peter Gentsch (Co-Founder and Senior Advisor of Retail.ai GmbH) and Norbert Hillinger (Retail & AI Associate) as well as representatives of leading industrial and retail companies, together with the experts Oliver M. Merx (lawyer and computer scientist) and Jan Berger (founder and CEO of Themis Foresight GmbH) on January 29, 2025 in Frankfurt am Main to discuss these developments and propose solutions.
The most important results and demands of the discussion round are as follows:
- Act instead of react: Early, pragmatic documentation to avoid or shorten costly audits and formal processes later on.
- Understanding regulation: The Retail.ai Academy offers comprehensive training courses that clearly illustrate the implications for retail and FMCG.
- Anticipate liability risks: A use case tool helps to identify and proactively manage the legal and regulatory implications of various use cases.
- Transparency and data sovereignty in generative AI: Recommendation on the use of open source solutions to build and promote an open and transparent AI community in retail.