According to TechSpot, the European Commission has opened a formal antitrust investigation into Google, probing whether the company “illegally” scraped online content to train its AI models. The case specifically targets Google’s use of web publisher content and user-uploaded YouTube videos to power generative AI services like AI Overviews and AI Mode, which are tied to its core search business. Officials are examining if Google breached competition rules by using this data without proper consent, compensation, or opt-out options, potentially giving itself an unfair advantage. The Commission’s vice president for competition, Teresa Ribera, stated the inquiry aims to balance innovation with legal principles. Google responded that the complaint “risks stifling innovation” in a competitive market. If violations are confirmed, the probe could result in penalties of up to 10 percent of Google’s annual global revenue, a figure that would amount to tens of billions of dollars.
The Core Issue: Scraping vs. Stealing
Here’s the thing: web scraping for search indexing is the foundational deal of the internet. Publishers let Google‘s bots crawl their sites so they appear in search results. It’s a symbiotic, if sometimes tense, relationship. But this new probe asks a much thornier question: did that permission for indexing extend to generative reuse? Basically, was it okay for Google to use all that crawled text and those YouTube videos as training data to build a product that might then replace the need to click through to the original source? The Commission seems to think that’s a different ballgame, especially if Google restricted competitors’ access to the same YouTube data trove. It’s the move from organizing the web’s information to digesting and regurgitating it that’s now under the legal microscope.
Why This Is Different
Now, this isn’t being done under the shiny new Digital Markets Act (DMA), which has its own set of rules for “gatekeepers.” Instead, the EU is using its classic, well-worn antitrust rulebook. That’s significant. It suggests regulators are framing this not just as a potential breach of a new tech-specific law, but as a fundamental abuse of market power under traditional competition principles. They’re arguing Google might have used its dominant position in search and video (YouTube) to hoard the training data needed for the next generation of search, locking others out. So it’s less about the act of scraping itself, and more about how market power dictates who gets to scrape what, and on what terms. That’s a much broader, and for Google, potentially more dangerous, argument.
The Broader Data Reckoning
This isn’t just a Google problem. It’s the leading edge of a massive regulatory and ethical reckoning for the entire AI industry. Everyone—OpenAI, Meta, you name it—trained their foundational models on oceans of data scraped from the public web. At the time, it was the “move fast and break things” wild west. But now that these models are commercial products making billions, the question of where that training data came from, and whether its use was fair, is coming due. Publishers and creators feel used, and regulators are listening. This Google case could set a crucial precedent for what constitutes fair use in the AI age. Will it chill innovation, as Google warns? Or will it force a more sustainable, permission-based model for the next phase of AI development? The answers will shape the tech landscape for a decade.
