Meta deepfake safeguards inadequate, board says

Meta moved Thursday to remove two AI-generated videos targeting women after its Oversight Board said the company's safeguards for deepfakes were "consistently and fundamentally inadequate."

The board ordered Meta to remove the fabricated videos from Facebook and Instagram, saying the cases undermined company policies on hate speech, misinformation and harassment. Meta said it welcomed the decisions and would comply within seven days.

One video depicted a Labour Party councillor in Scotland making inflammatory comments about refugees. The board said there was no public record of her making the comment and that the video showed signs of fabrication, including mismatched audio and facial movement.

The second video targeted a Muslim woman campaigner in Europe, falsely depicting her eating junk food while giving mindful eating advice and doing absurd exercises in an effort to discredit her health advocacy.

Oversight Board co-chair Pamela San Martin said AI-generated deepfakes are increasingly used to harass and silence women in public life. The board's decisions in individual cases are binding on Meta, but its wider policy recommendations are not.

San Martin called for stronger labelling of deceptive content, tougher penalties for repeat offenders and removal of deepfakes that harass private individuals. Last week, the board urged Meta not to replace professional fact-checking with Community Notes in Latin America.