Les chiens errants sont toujours une menace en Roumanie. Un problème abordé avec indifférence par les citoyens, traité avec cruauté par les autorités, et qui permet à certains malins sans pitié de s'enrichir.
- Articles / Politique, Roumanie, Courrier des Balkans, animaux, Une - DiaporamaWritten by Mar Negreiro.
Artificial Intelligence (AI) is changing the way we look for information online. Search engines increasingly offer AI-generated answers that keep users on their platforms instead of redirecting them to external websites, and many are now using generative AI tools as search engines. This convergence has intensified competition between traditional search engines and generative AI platforms. At the same time, experts warn that such reliance may lead to informational dependency and a decline in users’ cognitive skills.
AI is shifting online search resultsThe transformation from traditional web search to an AI-driven model is becoming mainstream. Google, which accounts for about 90 % of the online search engines market, has added an ‘AI Overviews’ feature – an automated search-summary tool – to its search engine. The overview uses Google’s Gemini AI models to answer users’ queries based on sources selected by Google. According to a market analysis, about 60 % of informational search queries in Google now trigger the overview as the top search result. Google’s AI Overview competes against AI platforms like ChatGPT and Perplexity, as users increasingly use AI chatbots to search for information. Similarly, Microsoft has introduced its Copilot generative AI to its search engine, Bing. However, Microsoft Bing accounts for just around 4 % of the online search engine market.
According to experts, users are less likely to click on links when Google’s AI summary is provided. They also believe that users are now structuring search queries differently and are increasingly using natural language in voice commands. AI answers are the latest feature of search engines, after voice and image queries.
Main emerging challenges Lower traffic and less revenues to publishersTrade associations representing publishers’ interests claim that AI overviews cause a drop of up to 25 % in publishers’ traffic, or even up to 50 % according to a digital marketing company. A network of digital marketing companies reports that 58 % of Google searches now end without the user visiting a single link. Reduced traffic translates into fewer ad impressions and a subsequent loss of revenue for content creators. In response, a coalition of publishers has filed a formal antitrust complaint with the European Commission against Google, accusing its ‘AI Overviews’ of diverting traffic and revenue from them. An alliance of German media and digital industry NGOs, associations and organisations has filed a formal complaint in Germany. A third case launched in the UK confirms Google has strategic market status in search services, including AI overviews and AI mode, the latter being its AI chatbot launched to compete against other chatbots.
The plaintiffs believe AI overviews are increasingly monetised through integrated advertisements, while Google extracts content from third-party websites without offering direct compensation or opt-out options, and without obtaining prior consent. They claim this has implications for media diversity, freedom of opinion and democratic discourse. Representatives of publishers’ interests claim they cannot opt out from appearing in the overviews without also being suppressed from the regular search listing.
Difficulties in the training of AI modelsTools like Google’s AI Overviews rely on two components: a large-language model (LLM) and a system able to retrieve information. Both need data, the first to learn to use human language and the second to retrieve documented information to answer users’ requests. Users and businesses may have grown accustomed to their data being collected and used for search results or targeted advertising, but the same may not be true for AI training. Complaints in the United States (US) point to an update of Google’s privacy policy, adding that the company may use publicly accessible information to train its AI tools. In fact, training LLMs with data publicly available on the web may raise copyright issues. Similarly, there is a case against Meta for not respecting the EU General Data Protection Regulation (GDPR) when collecting data to train their model, as users were not asked for prior consent to their data being collected for this purpose. However, recently proposed amendments to the GDPR might clarify that processing personal data for AI training may be permissible. Stakeholders are challenging Google. For instance, Cloudfare, a global platform and network of data centres that maintain websites and applications, is supporting a consortium of publishers to collectively block AI crawlers from collecting content for training unless companies pay for the content.
Threat to the open webUntil recently, Google drove online traffic in the open web, the public, interconnected network of websites and online services used for browsing, shopping and consuming news. Thus, Google has helped users find information by crawling and indexing web pages and ranking them based on a number of criteria. This is changing now, as users are offered generative AI-based responses and conversations and increasingly stop clicking and driving online traffic. Google has admitted that the open web is in rapid decline, referring to open-web display advertising. However, many believe this decline is much broader, as AI-driven search poses an unprecedented threat to the open web. In addition, users are increasingly using chatbots to conduct online searches, further reducing the number of users who access websites from search engines.
Risks for information quality, critical thinking and digital inclusionAI-generated searches may include incorrect or biased information. Research suggests that AI could produce ‘chat chambers‘ that reinforce the misinformation it hallucinates. Thus, 27 % of US adults do not trust AI-generated search results. Concerns around AI-generated misinformation are high, with the likelihood of chatbots repeating false information nearly doubling from 18 % in August 2024 to 35 % in August 2025. In addition, higher confidence in generative AI is associated with less critical thinking, according to research. It could also widen the digital divide, as young users are more AI-literate than older ones, but young students are increasingly dependent on AI use in classwork, risking lower creative thinking.
Next stepsGoogle’s AI Overviews are already available in over 200 countries worldwide. However, within the EU, rollout has been slower. In early 2025, the feature launched in only a few Member States; by October 2025 it was available in all except France. The delay was due to regulatory reasons. It is to be seen if Google is fully complying with key EU level legislation, including the Digital Services Act (DSA), the Digital Markets Act (DMA), EU copyright rules and the AI Act. The Commission is currently assessing Google’s use of AI summaries at the top of its search results, given their potential to significantly disrupt the relationship between Google and the open web by reducing traffic to external websites. Moreover, AI integration now goes far beyond overviews and chatbots. AI tools are increasingly embedded in operating systems, core apps and hardware ecosystems. As a result, AI has become a driving force for digital services, with further automation expected through the integration of Agentic AI or virtual assistants – systems designed to make decisions with minimal human oversight. Google, alongside other big tech companies, such as OpenAI, Microsoft and NVIDIA, is developing such systems as the next step towards automated online search, without any human involvement and replacing online search engines like Google Search. There is also a risk that AI agents could themselves become ‘gatekeepers’, as defined by the DMA, acting independently and creating new questions around oversight and accountability.
Read the complete ‘At a Glance note’ on ‘Search engines in times of Artificial Intelligence‘ in the Think Tank pages of the European Parliament.