Meta’s independent Oversight Board ruled on November 25, 2025 that manipulated videos can remain on Facebook with prominent “High-Risk” labels rather than automatic removal. The decision stemmed from a viral clip falsely depicting global protests supporting former Philippine President Rodrigo Duterte during his ICC extradition crisis. The video, viewed 100,000 times, used miscaptioned Serbian protest footage overlaid with pro-Duterte chants and the Filipino protest song “Bayan Ko.”
Case Background and Initial Handling
Meta’s automated systems flagged the post as potential misinformation, reduced its visibility for non-US users, and queued it for fact-checking. High queue volume prevented review despite similar videos rated false by Philippine fact-checkers. A user reported the reshared content; Meta upheld keeping it online, leading to Oversight Board appeal.
Board’s Legal Findings
The Board agreed the content legally stayed up, falling outside Meta’s civic misinformation policy prohibiting false voting information, dates, or procedures. However, it warranted “High-Risk” designation for digitally altered, photorealistic media posing deception risk during public crises. MIT research shows false news spreads 70% faster than truth.
Comprehensive Systemic Recommendations
- Publicly document manipulated media labels and application criteria
- Create separate fact-checking queue for similar-to-fact-checked content
- Prioritize identical/near-identical reuploads during political crises
- Equip fact-checkers with viral detection and rapid identification tools
- Apply all enforcement options consistently across misinformation campaigns
Meta’s Existing Framework and Industry Context
IFCN-certified fact-checkers typically label and downrank false content rather than remove. Proposed “High-Risk” tier aligns with C2PA metadata standards and Content Authenticity Initiative. YouTube mandates synthetic media disclosure; TikTok requires labeling deceptive AI content during elections.
Election Cycle Execution Challenges
Global elections amplify synthetic video threats through recycled footage and minor edits enabling rapid reuploads. Success demands infrastructure for near-duplicate detection, timely prominent labels, fact-checker escalation paths, and share-time contextual warnings. Board’s transparency-over-removal approach tests Meta’s ability to make labels corrective rather than cosmetic.
The ruling balances free expression with harm reduction, prioritizing friction against viral deception campaigns without chilling legitimate political speech.
