The European Commission has escalated its scrutiny of Big Tech by launching a formal antitrust investigation into Google’s use of online content to train its generative AI models. Centered on concerns around fair competition and potentially unauthorized use of publishers’ and platforms’ data, the probe reflects broader regulatory anxiety around AI’s rapid proliferation. Announced on December 9, 2025, the investigation places Google’s Gemini AI suite under direct examination as regulators assess the company’s data acquisition practices, content partnerships, and market dominance strategies [CNBC, 2025].
Context: Europe’s Strategic Response to AI Consolidation
Google’s dominance in both search and advertising infrastructure has long been under the EU spotlight, resulting in over €8 billion in penalties since 2017. However, the stakes have intensified as generative AI promises new dimensions of market control. According to Margrethe Vestager, the European Commission’s EVP for competition policy, the AI-infused nature of Google Search and other services may create new gatekeeping dynamics that disadvantage content creators, particularly news publishers [European Commission, 2025].
Europe’s Digital Markets Act (DMA), enforced since March 2024, aims to prevent “gatekeeper” platforms from leveraging systemic advantages to suppress competition. The DMA explicitly covers AI functionalities embedded in core services. Regulators are concerned that AI may exacerbate platform entrenchment by enabling enormous-scale content processing without proportional content licensing. As of 2025, Google’s generative AI reportedly crawls and absorbs hundreds of billions of online documents to fine-tune its Gemini AI suite—raising questions about the consent, remuneration, and transparency afforded to content owners [Bloomberg, 2025].
What Exactly Is Under Investigation?
At the heart of the EU probe is whether Google has utilized content—particularly from European publishers—without appropriate licensing or compensation to develop Gemini’s foundation models. EU authorities are soliciting evidence from media organizations, competitors, and technologists on whether Gemini outputs, including summaries, answers, and AI-generated news, use proprietary publisher data without permission [Financial Times, 2025].
Additionally, regulators are probing self-preferencing behaviors, such as Google’s integration of Gemini AI responses directly into search results. If AI-generated answers reduce user traffic to organic publisher links, thereby undermining ad revenue and subscriptions, it could violate the DMA’s fairness mandates. A critical line of inquiry involves whether Google’s AI content strips attribution or monetization opportunities away from sources it was trained on.
Preliminary surveys, including one released by the European Publishers Council in November 2025, revealed that 38% of respondents had observed significantly reduced referral traffic from Google since Gemini updates were rolled into Search, suggesting a structural shift in how content flows through the platform [European Publishers Council, 2025].
Comparative Regulatory Momentum: Global AI Scrutiny Expands
The EU’s probe into Google mirrors a growing international wave of interventions into AI-enabled content usage. In the U.S., the Federal Trade Commission (FTC) issued warnings in October 2025 to several companies, including OpenAI and Google, over potential deceptive uses of online data in model training [FTC, 2025]. Australia’s Competition and Consumer Commission, meanwhile, is pushing for legislative updates to its media bargaining code, explicitly referencing generalized AI-derived news services [ACCC, 2025].
Notably, competition authorities are aligning more closely with copyright regulators to build cross-functional oversight of AI. The French Autorité de la concurrence and Germany’s Bundeskartellamt have independently announced joint studies on how foundation models reshape market dynamics for information access [Autorité de la concurrence, 2025].
What Google Has Said—and Not Said
Google has pushed back against allegations of unfair enrichment from publisher data. In statements following the EU announcement, the company emphasized that Gemini respects copyright law and that any training data use falls under exceptions applicable to public data scraping. The search giant also highlighted its recent licensing deals with select European publishers as evidence of “good faith negotiation” in an evolving regulatory field.
Despite these claims, Google has yet to fully disclose the scope of content ingestion for Gemini’s various deployment tiers. While Gemini Ultra, its flagship model, is marketed as trained on “licensed and publicly available” datasets, the precise balance remains opaque. Analysts at VentureBeat AI note that transparency remains limited by the absence of a shared framework for AI training data attribution—and that Google’s defenses may not withstand scrutiny under the backdrop of specialized DMA provisions [VentureBeat AI, 2025].
AI and Market Power: Compound Risks in Digital Ecosystems
Analysts have warned that Google’s ability to distribute AI functionality directly within Search, Gmail, YouTube, and Android devices amplifies not only market reach but also the impact of any competitive distortions. If Gemini consistently generates responses that reduce clickthroughs to third-party sources while being trained on those sources, it introduces a feedback loop that entrenches Google’s ecosystem control.
This phenomenon, called “AI moat cementing,” lowers discovery of competitor content while increasing user reliance on Google’s interfaces for knowledge synthesis—a dynamic documented by the McKinsey Global Institute in its October 2025 examination of generative AI economic power concentration [McKinsey, 2025].
Below is a comparison of three AI players, their model release timelines, and their known content licensing strategies:
| Company | Flagship Model | Licensed Content Partnerships |
|---|---|---|
| Gemini Ultra (2025) | Select deals with EU publishers; large-scale scraping cited | |
| OpenAI / Microsoft | GPT-5.5 (2025) | WSJ, Associated Press, Axel Springer |
| Anthropic | Claude 3.5 (2025) | Limited disclosed deals; governed by U.S. fair use principles |
As shown, Google lags behind OpenAI in publicizing large-scale licensing partnerships, increasing its vulnerability to regulatory interventions in jurisdictions with stronger content rights protections, like the EU.
Legal and Economic Implications for 2025–2027
If proven guilty of breaching DMA obligations, Google may face fines of up to 10% of its global annual revenue, potentially exceeding $20 billion. Additionally, the European Commission has the authority to mandate behavioral or structural remedies, including requiring separate disclosures of AI-generated content and URL-level attribution to training sources [Reuters, 2025].
Beyond financial impact, an adverse ruling could force Google to redesign how Gemini integrates into Search and Chrome, with potentially global technical spillovers. Legal scholars from the MIT Technology Review warn that harmful precedents could splinter AI development into “consent-driven” versus “scraped-at-scale” approaches, with the former raising barriers to entry for smaller players [MIT Tech Review, 2025].
From an economic standpoint, content-rich verticals such as news, reference, and rankings websites stand to regain leverage if regulatory frameworks enforce payment for data access. According to a November 2025 Gallup panel of European content creators, over 71% believe that AI regulation should mandate revenue-sharing from tech firms extracting value from copyrighted data [Gallup Europe, 2025].
Future of AI Regulation: Key Takeaways and Strategic Uncertainty
The EU’s investigation into Google’s AI content practices is not merely about retroactive accountability. It’s about shaping forward-looking AI governance. Antitrust, copyright, data privacy, and content traceability are converging into a single regulatory framework that could influence global standards.
Regulators in Canada and Japan have expressed interest in adopting DMA-inspired models to pre-regulate AI integrations into dominant consumer services. Meanwhile, investor pressure is growing for Google to proactively disclose training dataset origins and revenue-share schemes, with ESG funds highlighting content sourcing transparency as a 2025 KPI [MarketWatch, 2025].
In sum, the EU’s actions signal a coming inflection point in AI regulation. Whether Google—and its peers—can maintain innovation momentum while aligning with increasingly granular data consent models remains one of the central questions shaping digital markets through 2027.