Meta's deepfake rules are 'fundamentally inadequate', watchdog rules, ordering removal of Scottish Labour councillor smear video

From the Newshound archive · first published 19 September 2026.

Meta's Oversight Board has directed the company to pull two AI-generated videos targeting women, making it the first occasion the body has invoked hateful conduct rules to demand the removal of a deepfake. One fabricated clip falsely depicted a Scottish Labour councillor endorsing comments about refugee rape — remarks she called "contrary to everything she stands for" — and racked up 5,000 views before Meta intervened. The company had twice turned down removal appeals, arguing that no trusted partner had flagged the content and that it carried no electoral dimension.

The ruling marks the board's first use of hate-conduct provisions to compel removal of a deepfake, as distinct from prior cases involving sexual content.

A second deepfake falsely showed a Muslim health campaigner gorging on junk food while giving dietary advice, to discredit her work.

The board recommended nine policy changes, including algorithm demotion of high-risk AI content and mandatory click-through warning screens.