Business Leaders Sound the Alarm: EU AI Act Threatens Technological Sovereignty

(Diesen Artikel gibt es auch auf Deutsch.)

Top European business leaders sounded the alarm in an open letter to representatives of the EU's Commission, Parliament and Council of Ministers that the current draft regulation on Artificial Intelligence (AI) would "jeopardise Europe's competitiveness and technological sovereignty". On the other hand, the initiative does not effectively address "the challenges we are and will be facing". The massive criticism relates above all to the position that MEPs defined in mid-June and with which they are now entering negotiations on a final compromise with the member states and the Commission.

According to the authors of the appeal letter, there is a threat of setbacks, especially with generative AI systems, which can generate new texts, images, music or videos based on existing, often protected works and thus trained basic models. Especially European Foundation Models (competitors to models like the one behind ChatGPT by OpenAI) would be "heavily regulated regardless of their use cases". Companies developing and implementing such systems "would face disproportionate compliance costs and disproportionate liability risks". Such regulation could lead to highly innovative companies "moving their activities abroad [and] investors withdrawing their capital from the development of European Foundation Models and European AI in general". The result would be "a critical productivity gap between both sides of the Atlantic".

150 Signatories

The signatories of the appeal include over 150 executives from companies and associations such as Airbus, Bitkom, Burda, Deutsche Post, Dr. Oetker, Eon, FlixBus, Heineken, Holtzbrinck, Orange, Renault, Siemens and Viessmann. Timotheus Höttges and Ralph Dommermuth, the CEOs of Deutsche Telekom and the 1&1 parent company United Internet, have also signed.

"Like the invention of the Internet or the breakthrough of silicon chips, generative AI is the kind of technology that will be decisive for the performance capacity and therefore the significance of different regions," the authors highlight. "States with the most powerful Large Language Models will have a decisive competitive advantage." Chatbots would replace search engines, for example, "[by] establishing themselves as the assistants of our daily personal and professional lives". As a result, "they will also be powerful tools that shape not only our economy but also our culture". Europe could "not afford to stay on the sidelines".

The managers do not want to deny the "inherent complexity and challenges" that generative AI brings along, as well as the "undeniable need for proper regulation". Given the "profound impact AI has on many areas of life", it is essential to "properly train these models and ensure their safe use". Due diligence in model development, standard labelling of AI-generated content, and safety testing before new models are introduced to the market would have to be enforced. However, self-regulation by the industry was sufficient for this.

Anchoring such rules in law and providing them "with a rigid compliance logic" is, however, a "bureaucratic [and ineffective] approach" and ultimately does not fulfil its purpose, according to the appeal. European law should therefore "confine itself to stating broad principles in a risk-based approach". The implementation of these principles could be entrusted to a special regulatory authority composed of experts at EU level. What is important here is "an agile process capable of continuously adapting [the rules] to the rapid pace of technological development". Only in this way can Europe become "part of the technological avant-garde", on which the future of the continent depends. Building a transatlantic framework is also stressed by the appeal letter's authors and considered a prerequisite for credible safeguards to be put in place, and "to create a legally binding level playing field".

Aggressive Lobbyists

Earlier, for example, the co-signing German AI Association (Deutscher KI-Verband) had warned that the EU Parliament's course reflected "an unfounded and disproportionate fear of a new technology". AI Foundation Models are a basic technology, and the people's representatives had put them "under general suspicion in the sense of a high-risk application" and thus endangered their development in the long term.

"It is unfortunate that the aggressive lobby of a few is hijacking other serious companies," counters Dragoș Tudorache, one of the Parliament's negotiators. The draft AI Act provides for "an industry-led process to define standards" and a "light regulatory regime". As the Liberal underlined, "this requires transparency. Nothing else."

OpenAI has already successfully lobbied for key parts of the proposed regulation to be weakened and for the US company to have its requirements reduced, as the TIME magazine had reported exclusively after thorough investigations. According to a recently published lobby paper, the Microsoft partner proposed relevant amendments that the Parliament adopted in its line on the globally observed legislative initiative. According to the paper, Generative AI should not be classified as a high-risk technology from the outset. However, operators of Foundation Models would have to check them for foreseeable risks to health, safety, fundamental rights, the environment, democracy and the rule of law, and mitigate identified dangers if necessary.

(sih)