The Oversight Board is an independent adjudicatory body created by Meta (announced 2019, operating since 2020) to review appeals of the company's content-moderation decisions on Facebook, Instagram, and Threads, and to issue policy recommendations. It is funded by Meta through an irrevocable trust but is institutionally separate. Since 2025 it has been chaired by Paolo Carozza (Notre Dame Law; Pontifical Academy of Social Sciences).
Overview
The board reviews appeals of Meta's enforcement decisions and issues policy recommendations to the company. Its case decisions on individual content are binding on Meta, while its policy recommendations are not. Funding flows from Meta through an irrevocable trust, an arrangement intended to separate the board's operations from the company's direct control. Commentators have described the body as an experiment in private platform constitutionalism, that is, an appellate mechanism for a platform's speech governance, and it is frequently invoked, both favorably and not, in debates over how delegated, company-funded oversight might apply to frontier-AI lab governance.
Meta's enforcement is increasingly automated, and the board functions as an external review point on the individual-case errors of those systems (AI and Content Moderation). The connection to AI-governance debates also runs through its chair: Carozza is among the more prominent Catholic-intellectual voices on AI governance, and his membership in the Pontifical Academy of Social Sciences (PASS) links the board to the Vatican-adjacent governance thread around Pope Leo XIV.
Account-level enforcement
In its first account-disabling case, decided June 4, 2026, the board upheld Meta's permanent disabling of a widely followed Instagram account that had shown a pattern of repeated violations, including visual threats of violence. After reviewing hundreds of public comments, the board found systemic due-process and human-rights concerns in how Meta disables accounts and governs account-level enforcement, citing opaque procedures, inadequate notice to users about what triggered enforcement, and weak appeal avenues. It recommended clearer rules and better user notification (Source: oversightboard.com).
The case extended the board's review from individual content decisions to account governance, including penalties, disabling, and recidivism systems. Account-level enforcement is the area where Meta's automated systems operate with the least transparency and the highest stakes for affected users.
Report on LLMs and political speech
In a July 2026 report, "Are LLMs Stifling Political Speech?", the board assessed how AI models handle political expression, testing 10 LLMs from six providers. It found that the models self-censor political criticism of leaders in restrictive jurisdictions — refusing to produce pamphlets critical of Xi Jinping or Mohammed bin Salman while producing equivalent material about Donald Trump and King Charles III — and recommended jurisdiction-specific geo-blocking rather than global application of national speech rules (Source: oversightboard.com). Fortune's coverage, beginning July 16, framed the finding as evidence that chatbots could function as an effective propaganda channel when national speech restrictions shape model behavior globally (Source: fortune.com). The report extended the board's remit from Meta's enforcement systems to industry-wide assessment of generative-AI speech governance. See AI Political Bias, AI and Authoritarianism.
Funding
Reversing earlier signals that funding might end after 2028, Meta announced on May 28, 2026 that it is increasing funding to the board. Casey Newton of Platformer described the announcement as "a new lifeline" (Source: platformer.news). The reversal followed the co-optation narrative that had attached to Carozza's chairmanship during the May 2026 encyclical coverage. Meta's treatment of the funding decision is covered on Meta AI.
Relationships
- regulated-by: Meta AI (the board adjudicates Meta's enforcement; Meta funds the board)
- related: Paolo G. Carozza (chair), AI and Content Moderation, AI Power Concentration, Algorithmic Accountability and Bias Audits
- related: Pope Leo XIV (via Carozza's PASS membership)
Sources
3 supporting sources (Oversight Board announcement June 4, 2026; Platformer May 28, 2026; board background). Created by the 2026-06-05 gap scan: 3 inbound wikilinks from Meta AI and Paolo G. Carozza with no page behind them, plus the live June 4 first-accounts-case thread.