Wikipedia has paused a trial that used artificial intelligence to write short summaries of its articles, following complaints from volunteer editors.
As per 404Media, the summaries were part of an experimental feature made available earlier this month to users who had a special browser extension and had chosen to take part. They appeared at the top of articles, but were hidden behind a click and marked with a yellow label reading “unverified”.
However, the test was met with immediate pushback from within the Wikipedia community. Editors said the summaries could mislead readers or contain errors, potentially damaging the website’s credibility.
One of the main concerns was the risk of AI producing inaccurate information — a problem often referred to as "hallucination", where the software invents facts or misrepresents them. Other publishers, including Bloomberg, have faced similar issues. Some have had to correct mistakes or scale back their own AI experiments as a result.
Wikipedia has said the feature is now on hold but hasn't ruled out the use of AI entirely. The Wikimedia Foundation, which oversees the platform, says it is still exploring how AI might help make the site more accessible — but insists any future tools must be accurate and trustworthy.
Image: DIW
Read next:
• Crypto Search Surge Places New York at the Forefront of U.S. Digital Currency Interest
• New WhatsApp Feature Summarizes Unread Chats Using Local AI, Bypassing Cloud-Based Data Handling
As per 404Media, the summaries were part of an experimental feature made available earlier this month to users who had a special browser extension and had chosen to take part. They appeared at the top of articles, but were hidden behind a click and marked with a yellow label reading “unverified”.
However, the test was met with immediate pushback from within the Wikipedia community. Editors said the summaries could mislead readers or contain errors, potentially damaging the website’s credibility.
One of the main concerns was the risk of AI producing inaccurate information — a problem often referred to as "hallucination", where the software invents facts or misrepresents them. Other publishers, including Bloomberg, have faced similar issues. Some have had to correct mistakes or scale back their own AI experiments as a result.
Wikipedia has said the feature is now on hold but hasn't ruled out the use of AI entirely. The Wikimedia Foundation, which oversees the platform, says it is still exploring how AI might help make the site more accessible — but insists any future tools must be accurate and trustworthy.
Image: DIW
Read next:
• Crypto Search Surge Places New York at the Forefront of U.S. Digital Currency Interest
• New WhatsApp Feature Summarizes Unread Chats Using Local AI, Bypassing Cloud-Based Data Handling
