The more than 110 companies led by ASML and SAP urge the European Union to postpone the implementation of the "AI Act": strict rules jeopardize the competitiveness of European AI.

European technology giants jointly urge the EU to postpone the implementation of new AI regulations. More than 110 institutions, including ASML (ASML.US), the world's largest chip manufacturer based in the Netherlands, European software giant SAP (SAP.US), and AI newcomer Mistral AI, recently wrote to EU Commission President von der Leyen, requesting a delay in the implementation of restrictive clauses targeting cutting-edge AI models. The open letter, jointly signed by European multinational companies such as ASML Holding NV ADR, SAP, Airbus, Mercedes-Benz (MBGYY.US), BNP Paribas, Lufthansa (DLAKY.US), and L'Oral, among others, stated: "Postponing the implementation in line with the commitment to regulatory quality will send a clear signal to global innovators of Europe's dedication to building a competitive environment." The core demand of the business community is focused on the lack of pace in rule implementation and the absence of detailed enforcement guidelines. According to the original plan, the strictest provisions of the EU's "Artificial Intelligence Act" were set to come into effect in August, requiring developers of cutting-edge models like ChatGPT to disclose training data details, establish copyright compliance mechanisms, and conduct additional risk assessments and architecture filings for high-risk systems. However, as of early July, a working group composed of scholars, developers, and advocacy groups was still discussing specific implementation guidelines, and key draft codes of conduct were significantly behind schedule compared to expectations in May. The controversy centers around the voluntary compliance framework proposed by the EU. Tech companies criticize the guidelines for having a suspicion of "overreach," with demands for third-party model audits and strict provisions for copyright tracing being considered too stringent. Meta's Global Affairs Chief Kaplan publicly criticized the current draft in February as "completely unworkable," and Alphabet executives stated that testing standards and intellectual property provisions far exceed reasonable boundaries. This regulatory uncertainty has already sparked a chain reaction: the United States government wrote to the EU in April urging a reevaluation of the feasibility of the legislation. As the world's first comprehensive AI legislation, the EU's "Artificial Intelligence Act" establishes a tiered regulatory system. Providers of basic models must disclose their technical pathways, high-risk applications (such as real-time facial recognition) face usage restrictions, and non-compliant companies may face fines of up to 7% of annual revenue, with a 3% upper limit for penalties for developers of cutting-edge models. While this "dual restraint" mechanism aims to prevent technology misuse, the high cost of compliance and ambiguous guiding principles are weakening Europe's attractiveness in the AI industry. The current implementation deadlock may force the EU to seek a new balance between regulatory strength and innovation vitality.