European companies have been swiftly implementing measures to label artificial intelligence (AI) systems, including chatbots and AI-generated images, audio, and video, as the European Union begins enforcing its AI Act’s initial consumer-facing transparency requirements. The rules, which came into effect on September 1, 2024, obligate companies to inform customers when they are interacting with AI technologies, marking a significant step in the EU’s broader regulatory framework for AI oversight.

The new transparency standards apply across sectors such as online retail, travel, and advertising, necessitating revisions to websites and customer service platforms. The European Commission now holds the authority to impose fines on businesses that fail to comply with these provisions, with penalties reaching up to 3 percent of a company’s global annual revenue or €15 million ($17.3 million), whichever is greater.

“This will be a cookie banner moment for AI,” said Patrick Van Eecke, a lawyer at Cooley, referencing the widespread adoption of cookie consent notices under Europe’s privacy laws. Until now, the AI Act was largely theoretical for consumers, but the new rules will make AI-generated content visibly identifiable.

Commission President Ursula von der Leyen has emphasized the importance of accelerating AI adoption within the EU to maintain the bloc’s competitiveness against global leaders such as the United States and China. The EU is also advancing plans for AI regulatory frameworks, increasing investments in computing infrastructure, and seeking to reduce bureaucratic obstacles to innovation. However, companies have expressed concerns about balancing rapid deployment with the complexities of new regulatory demands.

Berlin-based online retailer Zalando highlighted that the Commission’s recently published guidance, released just weeks before the requirements took effect, added “another unnecessary layer of regulation” and fell short of providing clear direction. Zalando cautioned that differing interpretations of the rules might lead to inconsistent compliance across the industry.

Similarly, retail trade group EuroCommerce warned against indiscriminate labelling, suggesting it could impose disproportionate costs and confuse consumers, while the lobby group CCIA Europe—representing major U.S. technology firms—criticized the broad scope of the deepfake definition, claiming it risks rendering labels meaningless if applied uniformly to all AI-generated content.

Advocates for transparency, such as Risto Uuk from the Future of Life Institute, argued that consumers deserve clear information about AI-generated content to understand associated risks. Commission officials have sought to address concerns about “labelling fatigue” by exempting certain minor AI modifications, such as standard photo editing, to ensure obligations remain practical and meaningful.

Online travel platform Booking.com noted that transparency aligns with customer expectations and can improve service outcomes. Pranav Pathak, the company’s AI product director, supported the regulation but called for clearer, straightforward guidance to help engineers implement the rules effectively.

The enforcement on September 1 represents only the initial phase of the EU’s AI Act implementation, which first came into force in August 2024. On the same date, the Commission gained expanded powers to scrutinize general-purpose AI systems deemed to pose “systemic risks.” Meanwhile, stricter requirements for so-called high-risk AI systems have been postponed until December 2027 following industry and member state lobbying.

Legal experts emphasize that businesses are challenged less by the legislation’s complexity and more by its evolving nature, making compliance unpredictable and difficult to manage in a dynamic regulatory environment.