Meta’s Oversight Board has criticized the company’s rules around manipulated media as “incoherent” and too narrow, following a review of a manipulated video targeting President Biden. The video in question depicted Biden seemingly making inappropriate comments about a young girl, but used AI editing techniques to manipulate his speech.
Oversight Board Orders Meta To Clarify Policies
The Oversight Board ordered Meta to clarify its stance and policies regarding manipulated media, stating that the current rules are unclear and fail to adequately protect users. Specifically, the Board took issue with Meta’s differentiated treatment of “cheap fakes” versus “deepfakes”.
- Cheap fakes are basic editing techniques like slowing down or splicing together videos
- Deepfakes use more advanced AI to create sophisticated and realistic forged videos
Under current policies, cheap fakes are permitted with limited fact-checking, while deepfakes have stricter rules. The Board argues this distinction is arbitrary and incoherent when the potential for harm is similar.
Edited Biden Video Can Stay On Platform
In a controversial decision, the Board determined the manipulated Biden video should remain on Meta’s platforms, despite concerns over misinformation. They stated that removing the video could set an overly broad precedent that would also sweep up “parodies and satire”.
However, the Board insisted Meta must add clearer labels noting the video is manipulated:
| Media Type | Required Label |
|---|---|
| Cheap Fakes | “Partially False Information” |
| Deepfakes | “Manipulated Media” |
This updated labeling is intended to curb the spread of misinformation by warning viewers of altered videos.
Broader Industry Issues Highlighted
Beyond just Meta’s policies, the Board’s review highlighted wider issues around detecting and handling manipulated media across the tech industry. They stated that while research is advancing, all companies still struggle with rapidly evolving manipulation techniques.
Some key industry challenges raised:
- Difficulty automatically detecting latest forms of synthetic media
- Balancing free speech versus platform integrity
- Preventing manipulated media from going viral during election seasons
Calls For Meta To Address Societal Impacts
The Board also urged Meta to better consider potential societal impacts from manipulated media and misinformation. Issues highlighted included:
- Truth decay eroding public trust
- Increased polarization and echo chambers
- Real-world harassment stemming from online hoaxes
There were demands for Meta to conduct more research into these areas and adjust policies accordingly, though the Board recognized the complex tradeoffs involved.
Next Steps
In response to the Oversight Board’s recommendations, Meta has 90 days to formally reply and clarify its stance. Key next steps include:
- Reviewing the arbitrary lines between cheap fakes and deepfakes
- Updating labeling rules around altered media
- Expanding researcher into societal impacts of misinformation
- Providing more transparency around efforts to detect synthetic media
Depending on Meta’s response, further policy changes may follow. The Board indicated it will continue monitoring Meta’s approach to these complex and evolving issues around manipulated content.
This latest ruling highlights the challenges tech companies face around misinformation, especially as improved editing techniques make falsified video content increasingly mainstream. However Meta chooses to update its policies, manipulated media will likely remain an issue through the 2024 election and beyond. Ongoing scrutiny from its Oversight Board and the public will pressure Meta to find solutions that balance integrity with freedom of speech.
To err is human, but AI does it too. Whilst factual data is used in the production of these articles, the content is written entirely by AI. Double check any facts you intend to rely on with another source.
