Meta’s Oversight Board Calls for New Rules on AI‑Generated Content, Urges Immediate Action

The board’s decision was triggered by an AI‑generated video that surfaced in late 2025, purporting to show bomb‑damaged buildings in Haifa during the Israel‑Iran conflict. The clip amassed more than 700,000 views before being reported, and Meta initially declined both to remove it and to attach a “high‑risk” AI label that would have signaled its synthetic origin. After the Oversight Board overturned that decision, it used the case to illustrate systemic gaps in Meta’s current approach, noting that the platform’s existing “AI Info” labels are “neither robust nor comprehensive enough to contend with the scale and velocity of AI‑generated content,” especially during crises (Engadget).

In its ruling, the board called for a dedicated policy that treats AI‑generated media as a distinct category from traditional misinformation. The proposed rule would spell out precise labeling requirements, define the circumstances under which users must disclose AI involvement, and outline the penalties for non‑compliance. By separating AI content from the broader misinformation framework, the board argues Meta can enforce more consistent standards and avoid the “incoherent” rule set it has previously criticized (TechCrunch). The board also urged Meta to move beyond its reliance on self‑disclosure and sporadic escalated reviews, which it deems insufficient for the rapid proliferation of synthetic media.

Detection technology was another focal point of the board’s recommendations. It demanded that Meta invest in “more sophisticated detection technology that can reliably label AI media, including audio and video,” and expressed concern that the company’s own watermarking of AI‑generated output is applied inconsistently (Engadget). The board’s emphasis on digital watermarks reflects a broader industry push for provenance signals that can be automatically verified by platforms and third‑party tools, a capability that remains under‑developed in Meta’s current ecosystem.

The board also highlighted the role of coordinated inauthentic behavior in amplifying deceptive AI content. In the Haifa video case, Meta ultimately disabled three accounts linked to the source after the board flagged “obvious signals of deception.” However, the board warned that without a separate AI rule, such networks can continue to exploit the platform’s existing misinformation pathways, muddying the distinction between genuine news and fabricated footage (Engadget). By mandating clearer labeling and stronger enforcement, the board aims to give users the ability to discern real from synthetic, particularly on matters of public interest.

Meta has 60 days to formally respond to the board’s recommendations, a deadline that underscores the urgency the Oversight Board places on curbing deceptive AI‑generated content. The board’s critique follows earlier condemnations of Meta’s “manipulated media” rules, which it has labeled “incoherent” on two prior occasions (TechCrunch). If Meta adopts the board’s suggested rule set and upgrades its detection and watermarking infrastructure, it could set a new benchmark for large‑scale social platforms grappling with the tidal wave of AI‑driven misinformation.