BRUSSELS — The European Commission’s chief AI adviser told the bloc’s 27 commissioners on Sept. 4 that Europe should focus less on winning the race to build artificial intelligence and more on using it to lift the bloc’s sluggish productivity. That’s the Brussels version of progress: keep the machine moving, keep the markets happy, and call it strategy.
Jim Hagemann Snabe, the EU’s new special envoy for industrial AI, told commissioners at a private seminar that the technology is Europe’s best chance to accelerate growth in health, agriculture, transport, defense and manufacturing, according to two Commission officials who were granted anonymity to discuss the briefing. The language is familiar. AI as growth engine. AI as competitiveness fix. AI as another tool for the same institutions that have spent years treating social life as a spreadsheet.
The Brussels Growth Machine
Commission President Ursula von der Leyen said in her State of the Union speech last week that Europe should not worry too much about leading the development of frontier AI models, but should “be the ones who draw the greatest value from it” while managing its dangers. That’s the real pitch. Not democratic control. Not public ownership. Value extraction, dressed up as prudence.
The EU is trying to capture the economic gains of rapidly advancing AI even as it grapples with the technology’s risks to jobs, security and society. The bloc’s institutions present themselves as referees, but the whole setup is built to serve the same old priorities: productivity, competition, and the smooth operation of capital. The people most exposed to the fallout are the ones expected to adapt.
Rules for the Marketplace, Not the Marketed
Article 50 of the EU AI Act, which became applicable in early August, requires AI-generated or modified content to be labeled. The EU classifies an image of an apartment that is “furnished using AI” as partially AI-modified and requires the photos to have an AI transparency icon. The providers of AI content, such as Open AI or Anthropic, must label images in a machine-readable format. A deployer, someone who uses an AI system to create or enhance an image, must also include a visible label on the image.
Claudio Teixeira, head of digital policy at the European Consumer Organization, said AI deceit in online marketplaces, such as the housing industry, are not talked about enough. “Not everyone has the capability to see at first glance that this is indeed not real,” he said. “So this content needs to have a clear warning that it is AI generated.” That’s the bare minimum, and even that arrives after the damage is already baked into the platform economy.
David Rappenglück, a CMS senior law associate who monitors applications of the act, said, “Article 50 is more or less an expression of this protection of fundamental rights.” He added, “The message is basically quite easy: if you, as a customer or a bidder, are confronted with a picture which might be AI generated or manipulated, you should be informed about it.” The wording sounds tidy. The reality is messier. The same digital systems that sell homes, status, and access are now expected to police their own tricks.
Akvilė Šikšniūtė, who works on the digital policy team in the European Consumer Organization, said consumers still face uncertainty. “I think we're still quite far away to the point where the consumers feel that they have places to turn to if they feel like they're misled or the content that they're seeing is not accurate,” she said. “I feel like this is where perhaps the misconception comes from for consumers slightly, because we expect now from all the communications we see that all the content will be labeled, and that's not the case.”
She also said the act does not distinguish between partial manipulation and full generation. “It isn't clear because, for example, very subtle changes like changing a background wall could be considered as a minor change that AI was used to change that. Therefore, it perhaps would not even be marked,” she said. The law arrives with icons and categories, but the market keeps its usual advantage: opacity.
Enforcement, Fines, and the State’s Favorite Word
Those who do not follow the act could be fined up to €15 million or 3% of their worldwide turnover. Rappenglück said, “We as lawyers would really recommend taking this quite seriously.” He added, “There's quite a hard enforcement mechanism and it's also enforced by the national authorities.” The national authorities, of course, are the same ones that will be asked to make the system look orderly after the fact.
Because the act only came into force last month, there are still a lot of unknowns about how it will work, and enforcement remains difficult in practice. Rappenglück said images can be screenshotted and reported to the national government, but real estate listings may only be up for a few days, making it harder for individual cases to be monitored. Teixeira asked the obvious question: “How do you actually ensure accountability of this sector to make sure that these practices do not gain hold? Right now, the apparatus is clear that this is not happening.”
The Professional Institute of Real Estate Agents in Belgium is currently updating its code of ethics to align with the rules of Article 50. Steven Lee from the institute said a real estate agent remains responsible for what they publish and can use AI only if they disclose it. “They have to follow the code of ethics to be able to exercise their activities. And if they don't, I'm afraid they can get into quite a bit of trouble,” he said. That trouble could include suspension or revocation of their license. The system knows how to punish. It just moves slower when ordinary people need protection.
The European Commission is also prepared to defend a €120 million fine against X after the US Justice Department said it supported Elon Musk’s effort to have the case annulled. The Commission fined X late last year for breaches of the bloc’s landmark digital protections, after a two-year investigation. It was the first penalty issued under the bloc's digital regulations governing online content. Regulators said X’s paid-for blue checkmarks violate the EU’s online transparency obligations.
The European Commission said in a press release at the time: “On X, anyone can pay to obtain the ‘verified' status without the company meaningfully verifying who is behind the account.” Commission spokesperson Thomas Regnier said Thursday that it is the EU’s “sovereign right” to draft legislation aimed at protecting its citizens. “We are enforcing our legislation objectively, transparently and with a solid case,” he told reporters. “We are ready to defend our position in court. We have a lot of evidence at our disposal, and it will be for the court to decide, as always.”
The commission is the top enforcer of the 27-nation EU's Digital Services Act, which requires tech companies to increase transparency, strengthen user protections and give audiences greater control when they navigate online platforms or face hefty financial penalties. The rulebook has become a flashpoint with the Trump administration, and US officials have criticised it as amounting to online censorship. Assistant Attorney General Brett Shumate said the “European Commission inappropriately attempted to expand its regulatory authority to reach American companies not present or operating within its jurisdiction.” He added, “We will not tolerate the European Commission engaging in regulatory overreach to try and control American engines of innovation and economic growth.”
The Justice Department warned that the fine, if upheld, “may have significant implications” for US online platforms and other companies providing digital services in the EU. When asked whether the Justice Department had informed the commission that it intended to back Musk, Regnier said: “We do not need to be informed about anything.” He said the commission doesn't expect the case to hurt broader relations between Brussels and Washington, notably in talks to mitigate the damage of Trump administration tariffs. “From our perspective, nothing will change because of a DSA case,” he said. The machine keeps its diplomacy. The platforms keep their leverage. And the people caught inside the system are still expected to trust the labels.