After the launch of ChatGPT in 2023, Wikipedia saw a steady increase in articles that volunteer editors had filled partly or entirely with text produced by chatbots. To curb this trend, the English-language community deployed a tool called Pangram that uses artificial intelligence to identify likely machine-generated content.
When Pangram flags an edit as suspicious, Wikipedia does not automatically treat the editor as a cheater — the organization notes that AI can make mistakes. Editors whose contributions trigger the filter receive a letter informing them that their edit was detected by the system. Following the deployment of these measures, the number of flagged articles has fallen significantly from a peak recorded in spring 2025.
Key figures about Wikipedia
- Since its founding 26 years ago, the nonprofit encyclopedia has grown to host content in 348 languages and contains more than 65 million articles.
- The English-language community is the most active, with about 115,000 contributors.
- The Hungarian Wikipedia contains over 570,000 articles and currently has roughly 2,800 editors.
Rules and practical guidance on AI use
Since spring 2024, Wikipedia contributors have been required to follow stricter rules when using AI. Principal points include:
- It is forbidden to create or comprehensively rewrite articles using artificial intelligence.
- Small edits produced by AI are allowed, provided a human reviewer checks them afterwards to avoid introducing unverified content or AI "hallucinations."
- Translating articles from other language versions using AI tools is permitted, but translations must be thoroughly reviewed before publication, with special attention to cited sources and references because AI often invents such details.
At the same time, Wikimedia leaders encourage editors to use AI in supportive roles — for example, to locate sources, suggest new topics, or request feedback and assessments of completed articles.
Why this matters
Wikipedia’s reliability depends on verifiable, sourced information. The rapid spread of AI tools presents new risks: convincing but false claims can be inserted more easily. Pangram and the tightened policy framework aim to protect the encyclopedia’s quality while still allowing AI to play a helpful, supervised role.
Looking ahead
The Wikipedia community continues to monitor developments and refine both detection tools and governance: the approach is not to ban AI outright but to regulate its use tightly and permit detection technologies to assist community moderation.
