Wikipedia volunteers build tools to detect AI-generated content

By Splitfeed EditorialLast updated Oct 21, 2026, 13:11 UTC

WikiProject AI Cleanup is a volunteer group on Wikipedia founded in 2023 by Ellia Lebleu and others to find and remove AI-generated text that violates the site's editorial guidelines. In an IEEE Spectrum interview published on 21 October 2026, Lebleu described how the project started when AI-written contributions were relatively easy to catch because they tended to rely on recognisable clichés. That phase of detection, however, is now largely over. Volunteers say the content has grown significantly more subtle, and the group is actively building new tools to keep pace with the change while also debating whether AI should have any sanctioned role in Wikipedia's editing process at all.

The stakes here go beyond a cleanup effort on a single website. Wikipedia sits at the centre of the open web's information ecosystem, feeding search engines, AI training datasets and countless reference tools. If machine-generated content embeds itself quietly into articles, the downstream effects on the accuracy of other systems could be substantial. The debate the volunteers are now having, about whether AI belongs in Wikipedia editing in some limited, approved way or should be kept out entirely, touches on questions about authorship, verifiability and editorial trust that no consensus has settled yet. The difficulty of detection is the detail that sharpens the dispute: when the problem was obvious clichés, the policy question seemed manageable.

None of the three camps have posted public reactions to this story, so what follows reflects the positions each would typically bring to a development like this. The pro-AI camp would likely argue that AI tools, properly governed, could assist volunteers rather than undermine them, speeding up routine editorial work and freeing humans for judgment-heavy tasks. The anti-AI camp would probably point to the growing difficulty of detection as evidence that the technology poses a systemic threat to open knowledge platforms that were never designed to handle content at this scale or with this level of opacity. The middle-ground camp would generally favour some version of disclosure and selective use, accepting that AI might play a support role while insisting on clear rules and human oversight for anything published.

The internal debate among WikiProject AI Cleanup volunteers about formalising AI's role, or rejecting it outright, is the immediate thing to watch. Any policy proposal put to the broader Wikipedia editorial community, or a statement from the Wikimedia Foundation on permitted AI use, would begin to settle at least the institutional side of the argument.