Wikipedia’s silent crisis: How AI is eating its own source 🤖

Cliquez ici pour lire en français

Wikipedia — the internet’s long-standing symbol of collaborative knowledge — is facing a new kind of storm. In 2025, its human traffic dropped by 8%, siphoned away by AI systems that can answer users’ questions instantly. The irony? Those same AIs are built on data pulled straight from Wikipedia itself.

Who’s stealing Wikipedia’s spotlight? 🎭

Over the past year, direct visits to Wikipedia have taken a noticeable hit. Between March and August 2025, AI-powered search engines and chatbots started delivering instant answers — often using snippets from Wikipedia without attribution. For users, that means no more clicking through; the information simply appears.

Meanwhile, younger audiences are turning to TikTok and Instagram for quick explainers and infotainment, further accelerating the shift away from traditional web browsing.

Bots on the attack 🕷️

Don’t write Wikipedia’s obituary just yet — but it’s fighting an uphill battle against a growing problem: large-scale data scraping. Armies of bots are harvesting content to train AI models, sometimes disguising themselves as real users to avoid detection.

This flood of fake traffic strains Wikipedia’s infrastructure and skews its analytics, forcing the Wikimedia Foundation to step up efforts to block malicious scrapers and protect the integrity of its traffic data.

Without Wikipedia, AI runs out of fuel 🛢️

Here’s the paradox: the very AI systems diverting users away from Wikipedia wouldn’t exist without it. Large language models rely heavily on Wikipedia’s accuracy, structure, and depth to generate coherent answers.

To make things easier, Wikipedia even provides structured data access for AI training. But the foundation laments that these platforms rarely give credit — or send users back to the source. Over time, that lack of reciprocity could weaken Wikipedia’s open model and its capacity to sustain free knowledge creation.

What’s next? A fragile balance 🚦

Wikipedia is exploring ways to integrate AI into its own tools — but that raises tough questions. How can it preserve the spirit of open, human collaboration in an ecosystem where AI depends less and less on the open web?

The issue isn’t just philosophical. Less traffic means fewer donations and a shrinking volunteer base, threatening the very model that’s kept Wikipedia alive for over two decades.

Pillar or stepping stone? 🏛️

The web is shifting toward a new reality: AI acting as the middleman between people and information. Everyone benefits from Wikipedia’s vast trove of knowledge — but if no one actually visits it, who ensures that knowledge remains accurate and up to date?

Finding a new, sustainable relationship between AI and open encyclopedias may be the key to preserving trustworthy, human-verified information in the age of instant answers.

Should Wikipedia demand compensation from AI giants — or reinvent its bond with readers? And how can we keep the world’s biggest encyclopedia alive in an era of one-click answers? Share your thoughts in the comments 👇