Did wikpedia / Wikimedia bow to the AI trend ?

If you’ve visited Wikipedia lately and felt a strange sense of déjà vu—or perhaps just a lingering smell of digital rot—you aren't alone. One of the last bastions of "human-curated" knowledge is currently under a state of siege so severe that the community has had to invent a new kind of digital yellow tape: The AI Maintenance Tag.

That screenshot isn’t just a warning; it’s a coroner's report. As of August 2025, Wikipedia formalised "Speedy Deletion Criterion G15," a panic button designed to nuke "AI Slop" before it can metastasize. We have officially entered the era where the world’s most famous encyclopedia is effectively a crime scene investigation, and you, the reader, are the unwitting witness.

The Hall of Mirrors: AI Citing AI

The irony is delicious, if you enjoy existential dread. Large Language Models (LLMs) were trained on the massive, human-written corpus of Wikipedia. Now, those same models are being used to "write" Wikipedia articles. We are witnessing the birth of a Self-Eating Information Loop. When an AI "hallucinates" a fact, dumps it into a Wikipedia article, and is then used to train the next generation of AI, the lie becomes the foundation. Facts aren't being discovered anymore; they’re being "statistically averaged" into existence.

The "Global South" Translation Scandal

It’s not just random trolls. Institutional "efficiency" is the real killer. Reports have surfaced of organizations using ChatGPT to mass-translate articles into dozens of languages. The result? Articles about European history suddenly featuring bizarre, hallucinated interludes about non-existent African deities or broken PHP code. The "translators" didn't check the work because, in the age of the Productivity Tax, checking the work is an expense no one wants to pay.

Institutional Despair: The Foundation’s Flirtation

While volunteer editors are working overtime as digital janitors, the Wikimedia Foundation itself has been caught looking at the "shiny new toy." Their briefly-tested (and widely hated) "Simple Article Summaries"—an AI feature editors called a "ghastly PR stunt"—showed exactly where the leadership's head is at. They’re terrified of losing traffic to Google’s AI Overviews, so they considered fighting slop with... well, more slop.

The Verdict: Verify at Your Own Risk (If You Still Care)

So, is Wikipedia unreliable now? Worse. It’s becoming a lottery.

To use Wikipedia today is to engage in a game of Russian Roulette with your own intelligence. Every time you read a paragraph, you have to wonder: Was this written by a subject matter expert with a passion for history, or was it shat out by a server farm in Oregon to meet a content quota?

The cynical truth is that we are moving toward a post-fact equilibrium. In this new world, "truth" is simply whatever prose was the cheapest to generate. Wikipedia isn't bowing to AI yet, but it’s drowning in it. The site has shifted from an encyclopedia you can trust to a digital landfill where you might—if you’re lucky—find a scrap of truth buried under a mountain of machine-generated garbage. Read at your own risk; the AI certainly didn't bother to write it for your benefit.