Meta Censured for Failing to Remove AI Deepfake Scam Videos
Meta's Oversight Board found that the platform left an AI deepfake video of footballer Ronaldo Nazário promoting an online casino scam active for months despite over 50 user reports. The board discovered approximately 3,900 similar fraudulent gambling advertisements using deepfakes of celebrities and recommended Meta strengthen human moderation of such content.
What happened
A deepfake video featuring Ronaldo Nazário was posted on Facebook in September 2024 advertising an online casino game called Plinko. The video combined Ronaldo's image with an AI-generated clone of his voice, misaligned in synchronization, promoting the casino with a link that redirected to a different game.
The video remained active despite receiving more than 50 user reports of fraud over months. The Meta Oversight Board found approximately 3,900 similar gambling advertisements on the platform, with about 3,500 containing deepfake videos involving the identities of other celebrities including Cristiano Ronaldo.
Reported impact
The video accumulated over 600,000 views before removal. The Oversight Board stated that the current moderation protocol put users at risk of financial fraud.
Index analysis
The Oversight Board determined the video was clearly a deepfake and should have been removed upon initial report. The board found that Meta failed to prioritize human review of the content despite repeated user reports, and that the post violated policies on fraud and deceptive practices.
Governance controls
The Oversight Board recommended that Meta strengthen its deepfake moderation policy by increasing human involvement and training staff accordingly. Meta responded that many of the board's allegations were 'simply imprecise' and committed to providing a full response within 60 days.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.