However, shortly after the forum, those who noticed these things noticed that US enthusiasm for regulating artificial intelligence appeared to have slowed significantly. It wasn't due to lack of interest in the forums or behind-the-scenes drama. Not because there wasn't a need. That was at least partially due to Dragos Tudrache.
Tudorache is a member of the European Parliament from Romania and chairs the Special Committee on Artificial Intelligence in the Digital Age. It is difficult to compare him with current American politicians, since his main character trait is a passion for seriousness. If you don't like white papers or long plenary sessions, close this tab and go directly to his Instagram. For the past five years, Tudrassche has been pondering how to regulate artificial intelligence in a way that prioritizes safety, shrugging off Europe's well-established reputation for fining Big Tech first and questioning it later. I've spent most of my time on that.
In September, Tudras visited Parliament House and briefed a group of four staff members on the European plan. Attendees described the session as “impressive” and “depressing.” The European Union's draft regulations are impressive because they are thoughtful, balanced, and flexible enough to continue to change in line with changing technology. That was also a frustrating part. While the US congratulates itself on starting the regulatory process, Europe is essentially done with it. And that set of rules was so good that Congress would soon have to choose between spending years trying to surpass it or copying the homework of clearly better students.
The decision point has arrived. The European Parliament overwhelmingly passed the AI Act last week, creating the first comprehensive set of rules on artificial intelligence. There are some details, but the people I spoke to from technology companies large and small, as well as civil rights groups, all agreed: “Copy your homework.” “I said to everyone I met in Washington, 'Listen, of course you think for yourself and decide what approach you take,'” Tudras said. “We have turned over the very stones that you are turning over now, and I think you will end up exactly where we are, not because we are smarter than anyone else. , simply because we started earlier than everyone else.''
As I said earlier, less prudent is a better way for Congress to save face. That's also strategic. Consistent AI rules across the free world would make compliance easier and cheaper for businesses and reassure a distraught public. The AI Act addresses the former by classifying AI based on a continuum of potential risks and building compliance to fit each category. If your AI product helps people organize their closets or recommend content, that's a very light operation. Whether an AI system sorts resumes, determines loan eligibility, or in the case of “fundamental” models like ChatGPT or Google’s Gemini, the model and its impact are regularly evaluated. It deals with people's fears by banning indiscriminate surveillance, social scoring systems, and anything else Chinese leader Xi Jinping can think of. “Regulations are more than just rules,” Tudras said. “This is an opportunity to express our values.”
The European Parliament has 705 members from 27 countries, and for decades its most important function has been as the continent's rubber chamber. Countries often sent their darkest political lights and most careless nepot babies to Brussels, where they could indulge in self-respect far removed from the real business of domestic politics. It's basically Fredo Corleone's strategy. Those days are long gone. While other governments around the world are becoming increasingly ridiculous, the European Parliament is often a beacon of competence and mature enough to learn from its mistakes.
The General Data Protection Regulation, passed in 2016, is a well-intentioned digital privacy law. It gives citizens the right to know how their data is used, the right to be forgotten, and forces companies to delete information under certain circumstances. It's also full of incomprehensible rules and vague enforcement responsibilities. It has made lawyers rich and companies doing business in Europe crazy. When Tudorache created rules simple enough for laymen to understand and advocated for a single enforcement authority, he had GDPR in mind. “I've spent thousands of hours talking with stakeholders from all directions and learning from everyone from Google to small startups,” Tudorache said. “Many of their process points made it into law, not because they lobbied, but because they made good arguments.”
I was silent for a moment when Mr. Tudras finished explaining how the AI industry does not lobby very openly. “I'm surprised,” he said with a little laugh, as if he had been waiting for this reaction. That's true. I didn't expect the Romanian-born European Parliament member to look so much like a Reagan-era Republican. If I didn't know better, I'd think he was wearing the Shard and expecting a Microsoft dinner.
If nothing else shakes Congress from its slumber, perhaps this will. There is no immediate risk that Europe will become the largest market for AI technology. The United States has a huge lead in terms of human resources, infrastructure, and capital. But if regulation is the love language of good governance, then Europe will continue to bravely protect individual rights while moving with the world's AI technology companies into a far more glamorous era. Check out his Instagram at Dragos Tudorache.