Last week, a group of technology companies wrote an open letter to European Union leaders urging the continent to fix its “inconsistent regulatory decision making” on artificial intelligence. The companies, Meta, Ericsson, and Spotify, wrote that “Europe has become less competitive and less innovative compared to other regions and it now risks falling further behind in the AI era.”
The letter focuses on open-source and multimodal artificial intelligence models, both priorities of Meta. Llama 3, the Facebook parent company’s top model, is one of the most powerful open-source models on the market.
Scott Wallsten, president of the Technology Policy Institute, was confused by the letter’s vagueness. Still, he pointed out that Meta was the only big name in AI to sign it, particularly because of the focus on open-source — no Microsoft, OpenAI, Anthropic, or Google. “They’re not saying that they oppose regulation. They’re saying that it’s inconsistent across the EU — and they don’t elaborate on that at all.” He said there are now three EU-wide regulations that will affect AI companies — the Digital Networks Act, the Digital Services Act, and the new AI Act — but the letter doesn’t clarify how they’re at odds.
But there are also hints that the critique is aimed at enforcement of the General Data Protection Regulation, or GDPR, the European data privacy laws. “In recent times, regulatory decision making has become fragmented and unpredictable, while interventions by the European Data Protection Authorities have created huge uncertainty about what kinds of data can be used to train AI models,” the letter says.
“The problem, as I understand it, is the GDPR doesn’t take a clear view on the legality of web scraping, so different EU data protection authorities [such as in the Netherlands and France] have each come to their own, slightly different conclusions,” said Nick Reiners, senior analyst for geo-technology at Eurasia Group, pointing at possible contradictions in the law. Plus, Reiners noted, Meta agreed to stop training its AI systems on EU data following a request from Ireland’s data protection authority. (Meta’s European headquarters, like those of many tech companies, are in Ireland.)
Reading between the lines, perhaps Meta wants to use data from Facebook, Instagram, and WhatsApp users in Europe to train its Llama models and have run into roadblocks in the law.
The letter is a pressure campaign with few specifics, a generalized complaint about how Europe has regulated largely foreign technology companies. If the signatories want real action, perhaps they should spell out where the laws clash and what can be done to fix them.