ChatGPT now faces stricter EU oversight as a very large search engine
The EU Commission has classified ChatGPT as a "very large search engine" under the Digital Services Act due to its built-in web search capability and over 45 million monthly EU users Reddit and Roblox were simultaneously reclassified as "very large online platforms" under the same regulatory framework All three services have until the end of December 2026 to comply with stricter oversight obligations, including risk assessments for illegal content, dangers to minors, and election interference A
Analysis
TL;DR
- The EU Commission has classified ChatGPT as a "very large search engine" under the Digital Services Act due to its built-in web search capability and over 45 million monthly EU users
- Reddit and Roblox were simultaneously reclassified as "very large online platforms" under the same regulatory framework
- All three services have until the end of December 2026 to comply with stricter oversight obligations, including risk assessments for illegal content, dangers to minors, and election interference
- A key legal ambiguity remains whether researcher data access rights under Article 40 extend to training data and model weights for systemic risk evaluation
- The EU Commission gains enhanced audit powers, working with Irish and Dutch authorities, bringing the total number of DSA-classified platforms and search engines to 28
Why It Matters
This marks the first time an AI chatbot has been classified under the DSA as a very large search engine, signaling the EU's expanding regulatory reach into generative AI systems. The ruling creates a precedent that could influence how AI-powered services are regulated globally, particularly around transparency, auditability, and researcher access. For AI practitioners, the unresolved question of whether training data and model weights must be disclosed represents a significant compliance uncertainty.
Technical Details
- ChatGPT's classification hinges on its web search functionality, which the EU Commission interprets as making it a "search engine" under the DSA, combined with its 45+ million monthly EU users exceeding the very large platform threshold
- Operational requirements include a publicly viewable ad archive, semiannual transparency reports, data access for approved researchers, reporting mechanisms for illegal content, a complaints process, and a crisis response mechanism
- The Commission gains audit rights over ChatGPT's risks and operations but does not have explicit authority to directly test the models
- Legal experts dispute whether Article 40's data access obligations extend to training data and model weights, with Professor Natali Helberger noting this will raise "interesting questions" about the scope of access necessary to identify systemic risks
- The EU Commission is coordinating with Irish and Dutch authorities for enforcement, reflecting the cross-border nature of DSA oversight
Industry Insight
- AI companies should proactively prepare for enhanced regulatory scrutiny by building internal compliance frameworks for transparency reporting, risk assessment, and researcher data access before enforcement deadlines arrive
- The ambiguity around training data and model weight disclosure creates both risk and opportunity—companies that voluntarily embrace transparency may gain regulatory goodwill, while those that resist could face prolonged legal battles
- This classification signals that the EU views AI search capabilities through the lens of existing digital services regulation rather than creating new AI-specific rules, suggesting a regulatory strategy of adapting established frameworks to emerging technologies
Disclaimer: The above content is generated by AI and is for reference only.