Wikipedia limits AI-written articles as AI summaries cut traffic

In short: Wikipedia is trying to keep human control over its articles as more AI-written text shows up and fewer people visit the site.

What's going on

Wikipedia has been preparing for a wave of AI-written content for years. In 2019, Wikimedia Foundation leader Katherine Maher warned that without humans, AI can cause serious problems.

Researchers and editors have since seen signs of AI-written text appearing on Wikipedia. A Princeton University study found that about 5 percent of nearly 3,000 new English Wikipedia articles in August 2024 had a substantial amount of AI-generated content. Editors also noticed patterns like unusual punctuation and leftover instructions from writing prompts.

Over time, many volunteers tried to fix the problem by removing generic filler text and fake sources. But the source describes a shift toward stricter limits. By March 2026, Wikipedia banned writing articles with large language models, which are AI systems that predict and generate text, like a very advanced autocomplete. Their use was restricted to translation and minor proofreading.

At the same time, Wikipedia is facing declining readership as more people get answers from AI summaries in search tools, such as Google AI Overviews. One report cited in the excerpt says Wikipedia lost almost a quarter of its readership since 2022. It also says people click links far less often from AI chatbots than from traditional Google search results.

What to watch

A key risk is a feedback loop. If human editors leave because AI content spreads, Wikipedia could be pressured to use even more AI to fill gaps, leaving fewer humans to check facts. For everyday readers, the big question is trust. If Wikipedia becomes less reliable, it gets harder to prove when confident-sounding AI answers are wrong.

Source: Wired