Oversight Board blasts ‘inadequate’ Meta safeguards for AI deepfakes

The Oversight Board has harshly criticized Meta's safeguards against AI deepfakes, deeming them "fundamentally inadequate." The board called on Meta to remove two AI-generated videos targeting women on its platforms. The first video depicted a Labour Party councillor in Scotland making false statements about refugees. There was no evidence of the councillor having made such remarks, and the video's audio did not match the timing of her facial expressions, indicating it was fabricated.

The second video targeted a Muslim woman campaigner in Europe, falsely portraying her as promoting junk food consumption while giving advice on healthy eating and performing bizarre exercises. The Oversight Board's co-chair, Pamela San Martin, stated that Meta's policies are "consistently and fundamentally inadequate" in addressing the surge of AI-generated content, particularly deepfakes used to harass and silence women in public discourse.

The board's decisions are binding on Meta, but its policy recommendations are not. Meta has pledged to comply with the board's directives within seven days. The Oversight Board has also urged Meta not to replace its professional fact-checking with Community Notes, a crowdsourced content verification model popularized by Elon Musk's platform, X. This request came after Meta began testing Community Notes in 16 Spanish-speaking countries in Latin America, as part of its transition away from Meta's fact-checking program, which operates in 26 languages.