Social Rocket
Wikipedia is caught in a paradoxical problem of AI search: its content is used massively by AI systems, while the direct relationship with people may weaken. Zach McCune, Global Head of Brand Marketing at the Wikimedia Foundation, describes exactly this tension to ADWEEK: Wikipedia is omnipresent in AI answers, yet the brand behind it is becoming increasingly invisible to users.
The Wikimedia Foundation had already reported 2025, after a revision of its bot detection, a decline in human Wikipedia page views of around 8 percent compared with the same months of the previous year. At the same time, automated access is growing strongly. For Wikipedia this is not just a traffic issue but a structural problem: fewer human visitors potentially also means fewer people who spot errors, add articles or later become editors themselves.
AI reach does not automatically replace brand relationships
For classic publishers, less search traffic mainly means fewer advertising and subscription opportunities. Wikipedia is funded differently, but faces a similar distribution question. When users get an answer directly in a search engine or chatbot, the original source easily fades into the background.
McCune frames this to ADWEEK as a brand problem: Wikipedia does not need to fight to appear in AI systems. The material is already there. The challenge is that people continue to know where the information comes from and who currently maintains it.
That is also why the 25-year campaign of the Wikimedia Foundation plays a bigger role than mere birthday communication. The organisation wants to make the people behind Wikipedia more visible and strengthen the connection between use, trust and active participation.
For brands and publishers, there is an important lesson here: visibility in generative answers is not the same as brand impact. The same problem now also concerns traditional media houses. Google is therefore trying, among other things, to make publishers more visible in AI Overviews and AI Mode through Preferred Sources.
Wikimedia Enterprise turns AI access into a product
Alongside its brand work, Wikimedia is building a technical and commercial response to AI demand: Wikimedia Enterprise delivers Wikipedia and Wikidata content via APIs for high retrieval volumes, search products, knowledge graphs and AI systems.
The Foundation names Amazon, Google and Meta, among others, as existing partners. Last year, according to an official Wikipedia 25 announcement, Microsoft, Mistral AI, Perplexity, Ecosia, Pleias and ProRata, among others, joined. Partners receive data at a speed and structure suited to their systems, thereby also supporting the non-profit infrastructure.
The model thus resembles a trend currently emerging among several publishers: instead of hoping that AI systems will keep classic referral traffic patterns, a separate market is emerging for machine-readable access to trusted content. Axios is now moving in a similar direction with Axios Direct, planning feeds for internal AI models and personal AI agents.
Google is also currently experimenting with usage-based payment for publisher content in AI answers.
Bots add to the infrastructure burden
Wikimedia also emphasises the technical strain caused by automated access. The Foundation reports that bot traffic has increased sharply and that scrapers for generative AI can cause considerable infrastructure costs. Services such as Firecrawl are therefore now switching from classic scraping to the official Wikimedia Enterprise APIs.
This underlines a second development: for content providers, the question becomes important not only whether their content is used by AI systems, but also how access takes place, whether attribution is preserved and whether the technical use is economically sustainable.
This boundary between useful agents and burdensome bot traffic is also relevant for marketing measurement. Automated access can distort analytics and retargeting if it is not cleanly detected, a problem we have already categorised in our article on AI agents and bot traffic in marketing.
What marketers can take away
Wikipedia shows particularly clearly where digital distribution could be heading. Content can appear in ever more systems and still generate fewer direct visits. For brands, publishers and creators, it therefore becomes more important whether their source, brand and relationship with the user remain visible in the AI answer.
The answer is unlikely to consist of SEO alone. Technical feeds, structured data, licensing models, direct platform partnerships and strong brand loyalty will grow in importance in parallel. Wikipedia has an advantage here: the organisation does not have to monetise every lost pageview. Even so, this example shows that even maximum AI presence does not automatically create a sustainable direct relationship with the audience.
Sources: ADWEEK on the Wikipedia Brandweek interview, Wikimedia Foundation on changed traffic patterns, Wikimedia Foundation on Wikipedia 25 and Enterprise partnerships, Wikimedia Enterprise.