This past March, the European Union was full of pride. After a lengthy process, the European Parliament had approved the AI Act, the first legislation specifically aimed at regulating the development and use of artificial intelligence systems.
European regulators’ intentions were probably good, as with the Digital Markets Act (DMA). In theory, both initiatives were born to protect the rights and privacy of European citizens from large technology platforms.
The problem is that the EU’s regulatory obsession negatively impacts European citizens, businesses, and developers.
They say it’s holding them back, that it’s leaving them behind. This is why.
We Have AI for Europe and Better AI for the Rest of the World
“We do not believe we will be able to roll out Apple Intelligence to our EU users this year,” Apple said last week. It was a big splash of cold water. As such, EU citizens hoping to enjoy the technology on their iPhones and MacBooks have to wait longer than expected.
Apple CEO Tim Cook. Image | Fortune Photo
Apple doesn’t want to take any risks and cites “regulatory uncertainties caused by the Digital Markets Act” as the reason for its decision. The DMA is very serious for big tech companies, and the one led by Tim Cook is quite familiar with it. On Monday, the European Commission opened another investigation into Apple over a possible violation of this new regulation.
In fact, this splash of cold water isn’t the first, and probably won’t be the last, that Europeans will receive regarding AI. The DMA and the AI Act make the future of AI very bleak for the EU. Although the intentions behind the laws are good—they aim to help avoid risks in the development of models and to protect the privacy of European citizens—the consequences are starting to be terrible.
We’ve already seen this with the mobile app version of Gemini, Google’s chatbot. It was available in 150 countries at launch, but none were in the European Union.
Google resolved the problem a month later. The same thing happened with Google Bard a year earlier, which took two months to be released in the EU.
The DMA and the AI Act make the future of AI very bleak for the EU.
In other cases, things are more serious. Microsoft announced Copilot for Windows 11 in September 2023. Nine months later, Europeans still don’t have official access to these features. A company spokesperson explained in April that “[the release of Copilot] has been delayed to 2024 due to EU regulations.” There are no definite dates for its availability, and Microsoft only adds that it’s working to make Copilot available “as soon as possible.”
One of the latest to be affected is Meta, which announced on June 14 that it would delay the launch of its Meta AI chatbot in the European Union. Regulators demanded that the company stop training its LLMs with Instagram and Facebook posts.
Meta AI isn’t available in Spain or other countries in the EU.
More advanced and experienced users can usually solve the unavailability problem. For example, they installed the Gemini mobile app by using its APK. In other cases, such as for chatbots on the web, it’s usually enough to use a VPN to “trick” the server into thinking users aren’t in the European Union. In the case of Copilot for Windows 11, there were also ways to skip restrictions.
The AI Act attempts to avoid risks in developing and deploying artificial intelligence systems. However, some experts warn this regulation could become a bottleneck for innovation.
That’s what Yann LeCun, Meta’s top AI manager, believes. LeCun told CNN that the big question about this legislation was the following: “Should research and development in AI be regulated? There are clauses in the EU AI Act and various other places that regulate research and development. I don’t think it’s a good idea.” For him, we’re far from the almost dystopian risks that European regulators are trying to avoid.
“Today, trying to figure out how to make future super-intelligent AI systems safe is like asking in 1925 ‘how do we make jet transport safe?’ And jet transport was not invented yet.”
"Being Too Cautious Before It's Time to Be Comes With a Cost"
To understand the problem better, we talked to Andrés Torrubia (@antor), an AI expert, entrepreneur, and co-founder of the Artificial Intelligence Institute (IIA) and Medbravo.
Torrubia began by recalling how the European Union’s theoretically good intentions can be counterproductive. This happened with the Cookie Directive. Protecting the rights of European citizens was a good idea, but how regulators implemented those protections was a disaster. Today, surfing the web is no better than before. It's (much) worse.
For him, “the impact of the AI law remains to be seen, and it's not known whether the cure will be better than the disease.” Torrubia says it makes sense to be cautious, but “being too cautious before it's time to be comes with a cost.”
European legislation has done just that, and European Union regulators have boasted that both the DMA and the AI Act are very protective of consumers and extremely good for innovators. Torrubia, an entrepreneur and an innovator, refutes this: “It’s a lie.”
“They say to themselves, ‘It’s not that bad. I only had one leg cut off, I can still walk,’ but you’re competing with a guy who has two muscular legs and is also on anabolic steroids.”
As he explained, “There is a short conquest phase in the technology world.” When a new market is discovered, “There are loopholes, and there is a big prize for being first. It's often achieved by exploiting gray areas.”
This is reminiscent of what happened with companies like Uber, Airbnb, Cabify, and LinkedIn. Today, you can’t create companies like the founders of those companies did. In addition, those companies have managed to dominate the market.
Torrubia said that some European entrepreneurs have a kind of Stockholm syndrome with this type of legislation.
“They say to themselves, ‘It’s not that bad. I only had one leg cut off, I can still walk,’ but you’re competing with a guy who has two muscular legs and is also on anabolic steroids.”
“What would I ask for to stimulate this strategic sector? I would ask for an unfair advantage that would put me on a par with the European market, the U.S., or even China. In return, I get something that holds me back.”
The late arrival of some AI options to the European market is an excellent example of this negative impact. At first glance, it doesn’t seem too bad that Europeans can’t access Copilot on Windows or must wait a few more months for the new Gemini or ChatGPT.
But in Torrubia's eyes, it’s serious. “You could have access a year later, and a group of young people would try it, get excited, start a company, and then be the seed of something else. However [because of the regulatory environment in Europe] their innovation would be hugely out of date.”
For Torrubia, the danger is clear. The cost could be colossal for a continent that, in his words, “lives on the success of the past.” Legislation like this, despite its likely good intentions, could end up condemning Europe.
In fact, it’s already doing so, to some extent.
Image | Rong_715 with Midjourney
View 0 comments