Whilst the EU AI Act was designed to protect the data rights of EU citizens, the net result is that EU citizens will not get access to the latest products from big tech companies, because those companies can no longer obtain EU data to train their models.

Organisations like Meta and Apple are taking a cautious approach with their latest AI products in the EU due to the regulatory requirements and potential liabilities imposed by the EU AI Act and other European privacy and data protection regulations, like the GDPR (General Data Protection Regulation). Here are some key reasons behind this hesitancy:

1. Strict Compliance Requirements

The EU AI Act introduces stringent obligations for high-risk AI systems, requiring extensive documentation, transparency and risk management protocols. For AI products that involve sensitive data (e.g., facial recognition, personalised recommendations, or generative AI tools that handle user input), organisations must comply with these requirements or risk significant fines and legal consequences.

2. High Financial Penalties and Legal Liabilities

Non-compliance with EU regulations can result in heavy penalties. For instance, GDPR fines can be as high as 4% of a company’s global annual revenue and the EU has already levied billion euro fines against Meta for GDPR infringement. With the EU AI Act, similar fines and penalties are likely, especially for organisations that fail to meet the new standards for high-risk or general-purpose AI models. These potential financial and reputational risks make companies cautious about launching new technologies without full regulatory clarity.

3. Transparency and Data Management Challenges

Many advanced AI products rely on large datasets for training, which may include personal information. The EU AI Act requires transparency in how these models are trained, the data sources used and how AI decisions are made. Additionally, AI providers must allow for human oversight and maintain logs for traceability. For companies like Meta and Apple, this translates into potentially extensive changes in how they handle and report data, increasing operational complexity. It could also require big tech companies to reveal more than they are willing in terms of how their ‘black box’ algorithms work.

4. Privacy and User Consent Concerns

Products that incorporate advanced features, such as facial recognition or location-based tracking, often trigger privacy concerns. European laws demand that companies provide clear information and obtain user consent for such data processing. Apple and Meta, for example, may prefer to limit their exposure to these regulations by withholding or modifying their services for EU users, avoiding conflicts with privacy watchdogs in the process.

5. Uncertain and Evolving Regulatory Landscape

With the EU AI Act’s framework continuing to evolve, companies face uncertainties about how the law will be enforced in practice, especially around the specifics of compliance for general-purpose AI models. Meta, Apple and others may decide to wait for more precise guidance and the establishment of standards to ensure their AI systems align with EU expectations before rolling them out to EU customers.

The EU’s regulatory environment demands high levels of transparency, data protection and accountability, creating significant compliance hurdles. While these regulations aim to protect users, they also increase the cost and complexity of deploying cutting-edge AI technology in Europe. As a result, companies like Meta and Apple are proceeding cautiously or limiting their AI offerings within the EU until they can be certain of compliance and manage associated risks effectively.