Oversight Board demands Meta overhaul AI-generated content rules after fake conflict video spreads

Oversight Board Challenges Meta's Approach to AI-Generated Content

Meta faces mounting pressure to fundamentally reshape how it handles AI-generated content across its platforms after the Oversight Board delivered a scathing assessment of the company's current policies. The 21-person independent board issued recommendations calling for dedicated rules separate from existing misinformation policy, improved detection technology, and more consistent use of digital watermarking 1

1

. The decision stems from a case involving a fake AI video that claimed to show damaged buildings in Haifa during the Israel-Iran conflict in 2025, which accumulated more than 700,000 views before Meta took action 2

2

.

The video was posted by an account in the Philippines masquerading as a news outlet. Despite multiple user complaints, Meta declined to remove it or apply a "high risk" AI label that would have clearly indicated the content had been created or manipulated with artificial intelligence 1

1

. The board overturned Meta's decision and disabled three accounts linked to the page after identifying "obvious signals of deception."

Critical Gaps in Detection and Labeling Systems

The Oversight Board's assessment reveals that Meta's current "AI Info" labels are "neither robust nor comprehensive enough to contend with the scale and velocity of AI-generated content," particularly during times of crisis or conflict where engagement spikes dramatically 3

3

. The board emphasized that a system overly dependent on self-disclosure of AI usage and escalated review cannot meet the challenges posed in the current environment 1

1

.

Meta's reliance on users to voluntarily label their own content and on fact-checking partners to flag problematic material has proven insufficient. The board noted concerns from these trusted partners that Meta "is less responsive to outreach and concerns, in part due to a significant reduction in capacities for Meta's internal teams" 1

1

. This approach means fake AI videos can spread rapidly before moderators catch them, particularly during major global events 4

4

.

Recommendations for a Comprehensive AI Content Rule

Among the board's top recommendations is the creation of a dedicated Community Standard for AI-generated content that operates independently from Meta's existing misinformation policy 3

3

. This new AI content rule should include specific details about how and when users are required to label content, as well as clear information about penalties for those who break the rule 1

1

.

The board also called for Meta to invest in more sophisticated, robust detection tools capable of reliably identifying AI media, including audio and video formats. Additionally, the board expressed concern about reports that Meta is "inconsistently implementing" digital watermarks on content created by its own AI tools 1

1

. The recommendations include better adoption of Content Credentials, an industry framework that attaches metadata showing whether AI tools were involved in content creation 3

3

.

Rising Threat of Deceptive AI Content During Armed Conflicts

The board described the 2025 Iran-Israel conflict as an "inflection point" for deceptive AI content, warning that the proliferation of AI content during armed conflicts has "challenged the public's ability to distinguish fabrication from fact ... risking a general distrust of all information" 2

2

. Since the start of US and Israel's strikes on Iran, there has been a sharp rise in viral AI-generated misinformation across social media platforms 1

1

.

A BBC analysis at the time found that fake AI videos posted after the conflict began—content that was either pro-Israel or pro-Iran—quickly collected at least 100 million views 2

2

. The board argued that Meta's current threshold for labeling AI-generated content is too high, particularly when the subject involves armed conflict. Meta had claimed the Haifa video did not require any label because it did not "directly contribute to the risk of imminent physical harm" 2

2

.

Meta's Response and Industry-Wide Implications

Meta has 60 days to formally respond to the Oversight Board's recommendations 1

1

. In its initial statement, the company said it would label the video at issue within seven days and would abide by the board's suggestions when it encounters "identical" content in the same context 2

2

. Meta also indicated it will implement changes "when it is technically and operationally possible to do so" 5

5

.

This isn't the first time the board has criticized Meta's handling of synthetic media. The group has previously described the company's manipulated media rules as "incoherent" on two occasions and criticized its reliance on third-party fact-checking organizations 1

1

. The board's suggestion that "the industry needs coherence in helping users distinguish deceptive AI-generated content" signals that these challenges extend beyond Meta to other platforms grappling with the rapid advancement of generative AI tools 1

1

. As deepfake videos become increasingly difficult to detect due to advances in AI technology that can now generate realistic lighting, motion, and voice synchronization, the need for proactive labeling of fake AI videos and stronger content moderation during crises becomes more urgent 4

4

.