Meta slammed for 'incoherent' deepfake policy after viral Biden video

In the lead-up to major elections worldwide, Mark Zuckerberg-owned Meta, the parent company of social media giant Facebook, is under increasing scrutiny after an oversight body declared an urgent need for updates to its policy on deepfake content.

The announcement, made on Monday (Feb 5), stems from a decision surrounding a manipulated video or deepfake featuring US President Joe Biden.

The controversial video, depicting Biden voting with his adult granddaughter, went viral last year after being manipulated to suggest inappropriate touching. Initially reported as hate speech, the incident was brought before Meta and later its independent oversight board.

Also watch |Oversight Board rebukes Meta’s policies on deepfakes, amid AI fears

The oversight board, responsible for reviewing Meta's content moderation decisions, acknowledged that the platform was technically correct in allowing the video to remain online. It also raised concerns about the adequacy of the company's rules regarding manipulated content.

The board's warning comes at a crucial time, with fears of widespread misuse of artificial intelligence-powered applications for disinformation on social media platforms during a pivotal election year, not only in the United States but worldwide.

Meta's current content moderation policy 'incoherent'

Describing Meta's current policy as "incoherent" and "lacking persuasive justification," the board called the need for a focus on specific harms, particularly those related to electoral processes, rather than on the content creation process itself.

In response, Meta stated that it is "reviewing the Oversight Board's guidance" and committed to publicly addressing the recommendations within 60 days. The board clarified that, in the case of the Biden video, the rules were not violated because artificial intelligence was not employed, and the video did not depict Biden making false statements.

However, the oversight board underscored that non-AI-altered content, often referred to as "cheap fakes," remains prevalent and can be just as misleading.

It pointed out that altered audio content, which can be highly effective for deception, is not currently covered by Meta's policy.

(With inputs from agencies)