The designation of ChatGPT as a Very Large Online Search Engine and the Geopolitics of the Digital Services Act

Michele Sacco
FoxTPNL, CC BY 4.0, via Wikimedia Commons

Introduction

On the 31st of August, 2026, ChatGPT was designated by the European Commission as a Very Large Online Search Engine (VLOSE) under the Digital Services Act (DSA). In the words of Henna Virkkunen, Executive Vice-President for Tech Sovereignty, Security, and Democracy, this designation means that ChatGPT “will now be held to a higher standard of scrutiny and accountability in the European Union, in line with [its] large impact on [European] citizens and society.”

However, the classification of ChatGPT as a VLOSE not only marks a major development in the enforcement of the EU regulatory framework but also in the awareness of the role played by AI in contemporary society. In light of recitals 75 and 76 of the DSA Regulation (the “Regulation”), the designation certifies for the first time that an AI Chatbot can play a “critical role in locating and making information retrievable online,” justifying stricter rules and a higher standard of due diligence obligations. 

Accordingly, the designation provides an opportunity to analyse how different emerging technologies and digital services can be caught under the DSA, as well as the extent to which the EU’s regulatory power is acquiring geopolitical significance.

Henna Virkkunen, Executive Vice-President for Tech Sovereignty, Security, and Democracy
© European Union, 2026, CC BY 4.0, via Wikimedia Commons
The designation of ChatGPT as a Very Large Online Search Engine (VLOSE)  

Without entering the legal conundrum of the rules and definitions of the DSA, it will suffice to highlight that the designation of ChatGPT as a VLOSE entails a reflection on both the nature and diffusion of the popular AI Chatbot. 

Firstly, the Commission found that ChatGPT functions as an ‘online search engine,’ which Article 3(j) of the Regulation defines as an intermediary service that allows users to input queries to perform searches of, in principle, all websites and returns results in any format in which information related to the requested content can be found. 

Secondly, ChatGPT met the threshold for an online search engine to be classified as a “very large” one, which Article 33 of the Regulation sets at 45 million monthly active recipients in the Union. While it is difficult to identify the exact number of ChatGPT users across the EU, the average number exceeds 100 million monthly users by far, triggering the VLOSE qualification. 

This decision marks a pivotal moment in public policy reflection on Large Language Models (LLMs). They are no longer seen as emerging technologies only capable of creating new content but as complex infrastructures that shape access to knowledge for many citizens as well.

Indeed, two “tectonic shifts” are revolutionising the way people conduct research online. The integration of generative AI summaries in traditional search engines and the rise of LLMs for research purposes have paved the way for the so-called “zero-click” search. In essence, while traditional online search engines respond to queries by providing a vast number of websites, LLMs and AI chatbots directly answer the questions posed by users. In other words, their algorithm plays a much more active role in selecting the sources and the information to extract from them, synthesising the content, and elaborating a complete answer which only provides very few links to external sources, if any. 

Now, even with traditional online search engines, very few people would have actually looked at the second page of results or even scrolled down to the bottom of the first one. However, the algorithmic power of LLMs in terms of access to knowledge is on a different scale. By serving as an entry point to information that reduces the plurality of sources actually consulted by users, LLMs present a series of potential risks that must be managed carefully, including the system’s mechanism to retrieve and select information, as well as its generative process in elaborating the answer.  

In short, after the designation by the Commission, ChatGPT has four months to adjust to the additional obligations set out by the DSA, which entail higher standards in terms of transparency, risk assessment mechanisms, compliance, due diligence, and collaboration with the Commission, as well as the exposure to higher fines in case of DSA violations.

Sandro Gianella, Anna Makanju, Jade Leung and Vĕra Jourová (Visit of OpenAI representatives to the European Commission)
© European Union, 2026, CC BY 4.0, via Wikimedia Commons
The main enforcement actions under the DSA so far

Although its adoption dates back to 2022, the DSA has seen increasing enforcement in recent times. Notably, since the end of last year, the Commission has been exercising the sanctioning powers that it is endowed with by the DSA in a series of high-profile cases. 

The first non-compliance decision under the DSA concerns Elon Musk’s owned X (former Twitter). The decision was based on three different violations. Firstly, the Commission found a deceptive use of X’s ‘blue checkmark’ for ‘verified accounts.’ By allowing anyone to pay and obtain the ‘verified’ status, without carrying out meaningful procedures to check the identity of the person behind the ‘verified’ account, X exposed users to scams, including impersonation frauds, as well as other forms of manipulation by malicious actors. Secondly, X fell short in several transparency obligations with relation to the accessibility of its advertisement repository and public data, impairing researchers and civil society in evaluating systemic risks in the EU and in detecting scams, hybrid threat campaigns, coordinated information operations, and fake advertisements. 

After imposing this first fine, the Commission turned its attention to two of the most popular Chinese e-commerce platforms: Temu and AliExpress. Overall, both companies were found in violation of similar due diligence obligations, especially with regard to the identification, analysis, and assessment of systemic risks of illegal, unsafe, or counterfeit products being offered on their platforms.

By the end of 2026, the Commission could go full circle and fine another major American social media company. This time, it is Meta to be in the crosshairs, and the emphasis is put on the allegedly addictive design of two of the most widespread social media platforms in the world: Instagram and Facebook. While a final decision has not been adopted yet, the Commission’s investigation indicates that Meta did not adequately assess the risks of its addictive design on the physical and mental wellbeing of users, including minors and vulnerable adults. Crucially, the Commission believes that Meta’s current mitigation measures failed to effectively tackle the risks stemming from its addictive design. 

ChatGPT availability by country or region
M.Bitton, CC BY-SA 3.0, via Wikimedia Commons
The Geopolitics of the DSA

The designation of ChatGPT as a VLOSE and the main enforcement actions outlined above testify to the regulatory power of the EU over emerging technologies and digital services. If one combines the “physiological” enforcement actions (such as ChatGPT’s designation as a VLOSE) and the “pathological” ones (such as the fines for non-compliance discussed above), it is clear that the DSA primarily targets foreign companies. This designation can be interpreted in different ways. 

A first approach would be to highlight the contribution that enforcing the DSA provides to the EU as a global actor. Leveraging its regulatory power, the EU is capable of pushing foreign companies to adopt European standards. This kind of analysis usually refers to the “Brussels effect,”  whereby the EU would eventually prompt other states to mirror the European regulatory framework. The negative reactions by foreign countries, especially by the current US Administration, would reinforce the thesis in favour of the geopolitical significance of the DSA and of other crucial pieces of EU legislation in the digital sector, such as the Digital Markets Act (DMA). 

However, one could question whether the over-politicization of EU law is positive or negative news. Indeed, pursuing geopolitical goals simultaneously strains the Commission’s credibility as an enforcer of EU law. Moreover, the Commission has yet to prove its ability to withstand increased pressure from foreign actors. Thus, it has been argued that the time has come to create an independent European Digital Authority, safeguarding the Commission from external pressure when dealing with other dossiers, such as international trade.  

Against this backdrop, the fact that foreign companies are the primary targets of the DSA already elevates its enforcement from technical operations to calculated geopolitical decisions. Yet, this does not prove the ability of the DSA to modify the behaviour of tech companies outside the EU. Accordingly, the geopolitical significance of the DSA must be interpreted narrowly: not as an immediate tool of foreign policy, but rather as an expression of European territorial sovereignty. 

This second conceptualisation of the EU’s normative power over the digital sector is again open to different interpretations. On the one hand, it demonstrates that the EU is not toothless vis-à-vis foreign tech giants. On the other hand, the absence of a competitive European technological industry should raise concerns about its long-term sustainability. For instance, looking at the list of the designated “gatekeepers” under the DMA, one realises that the primary addressees of such a crucial piece of legislation are not Europeans. Thus, while the European normative power is crucial to protect territorial sovereignty, the lack of European ownership in the digital sector significantly impairs the achievement of digital sovereignty, especially if foreign governments commit to shielding their companies from the application of EU rules or threaten widespread retaliatory measures.

Conclusions 

Classifying ChatGPT as a VLOSE is an important step toward enforcing the European regulatory framework in the digital sector. Moreover, it underscores the DSA’s positive role in governing emerging technologies and understanding their impact on society. For instance, the application of the DSA to ChatGPT should spark a deeper reflection on the power of AI in mediating between citizens and information. Similarly, the investigations into Meta should shift the focus from the content available on social media to the design and functioning of social media platforms as a whole. 

However, when it comes to the geopolitical role of the DSA, the conclusions are more somber. The enforcement actions underscore the European normative power and defend the Union’s territorial sovereignty. However, it is not through regulation alone that the EU will ever play a decisive role in the governance of the digital sector at a global scale.

Additional readings
Additional questions
  • To what extent should AI systems be responsible for providing unbiased, diverse, and reliable answers to socio-political questions?
  • Should the enforcement of the European digital normative framework be a purely technical matter, or should it accommodate political considerations as well?
  • Are central industrial policies necessary to achieve European digital sovereignty?

Leave a Reply

Your email address will not be published. Required fields are marked *

The designation of ChatGP…

by Michele Sacco time to read: 7 min
0