Europe comes for Big Tech, again
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Aïda Amer/Axios
European leaders have spent the week doing what they do best: regulating, and promising to poke around in other people's tech.
Driving the news: The EU's legislative reconciliation process is underway to finalize the bloc's AI Act. While officials met until nearly midnight — after a supermajority of legislators approved the text Wednesday — the process will likely take months, and the law won't take force until 2025.
- The EU's antitrust regulator on Wednesday issued new charges against Google's ad business, with the goal of forcing Google to break the business up.
- Google Tuesday was forced to delay the launch of its Bard chatbot in the EU, as the Irish data regulator demanded proof of EU General Data Protection Regulation compliance.
- U.K Prime Minister Rishi Sunak opened London Tech Week with the claim that he'd secured "early or priority access" to the models of OpenAI, Anthropic and DeepMind for the British government.
Between the lines: European parliamentarians have a strong interest in quickly finalizing the AI Act. EU elections take place May 2024, and AI guardrails are a popular measure, which handily demonstrate the EU's global influence.
- "Quality is just as important as speed," the Parliament's lead negotiator Eva Maydell told Axios from inside the negotiating room.
- EU digital enforcer Margrethe Vestager needs a win before her term ends in 2024, after a series of crushing court losses on major antitrust and digital tax cases.
- The U.K. government meanwhile is desperate to brand its post-Brexit economy as more innovation-friendly than the EU.
What's happening: The EU's AI Act lays out the first democratically negotiated rules for foundation models, and would require AI use cases to be categorized according risk, with some high-risk cases banned.
- Prohibited uses of AI could include emotion recognition systems, biometric databases built by scraping photos on the internet (such as Clearview AI), and real-time facial recognition — though diplomats are expected to insist on exemptions for national security systems in the final text.
- The British government is working to sell London both as HQ for global AI regulation, and as an AI innovation hub.
Be smart: The EU's two-year effort to craft AI regulation — which had to be patched at the last minute to account for generative AI uses — offers lessons for the U.S. and U.K., giving both the possibility of second-mover advantages.
Yes, but: While EU officials rail against facial recognition, U.S. Homeland Security agents have proved its use this month, identifying and arresting a man living in Tennessee on war crimes charges dating back to the 1990s Bosnian War.
Flashback: French, Italian and Spanish data protection agencies have all announced investigations into ChatGPT, and the EU's privacy watchdog announced a task force on the company's practices in in April.
