H1 2026 Report on the Oversight Board
UPDATED AUG 27, 2026
by Louise Turner
Today, we are sharing our H1 2026 Bi-Annual Report on the Oversight Board. This is our 16th report since the Board's inception, and it arrives at a moment when the questions the Board was created to answer have become central to a far wider conversation: how should content be governed in an era shaped by artificial intelligence? In six years of partnership, the Oversight Board has moved from reviewing individual decisions to shaping the frameworks through which platforms, regulators and the public think about accountability at scale.
This report tracks our implementation progress across the 347 non-binding recommendations the Board has issued through June 2026. We have committed to implementing, or exploring the feasibility of implementing, 80% of those recommendations, and have implemented 214 of them.
Some highlights from this past half include:
- Governing AI at scale, from provenance disclosure to industry research: The Board's recommendations in the Alleged Audio Call to Rig Elections in Iraqi Kurdistan and AI Generated Video in Iran-Israel Conflict cases have encouraged further work on provenance at content creation. Aligned with that direction, we publicly introduced Content Seal, our invisible watermarking technology that, alongside metadata, embeds provenance signals directly into AI-generated media at the point of creation. This year, the Board published two assessments including whether large language models from across the industry may be stifling political speech and also about where Meta should move faster on AI content. It also evaluated our approach to introducing more advanced AI into content moderation, and the Oversight Board is increasingly demonstrating how it serves as a model for what AI companies more broadly can learn from independent oversight.
- First review of a permanently disabled account: For the first time, the Board reviewed an account takedown decision, extending its impact beyond individual pieces of content to whether removing a person's account on our platforms was proportionate. The case drew more than 750 public comments, and rather than issuing standard recommendations, the Board set out principles to guide account governance across the industry. Piloting this expansion reflects the Board's growing institutional maturity and its ability to take on new forms of consequential enforcement questions.
- Advancing fairness and inclusion in enforcement: In the Emojis Targeting Black People bundle, the Board examined how hate speech can be communicated through emoji combinations rather than text, pushing our detection systems to evolve beyond their original text-based focus. Acting on this guidance, we deployed enhanced detection systems ahead of the 2026 FIFA World Cup calibrated to identify emoji-based hate speech across languages. Per a recommendation from the case on Criticism of European Union Migration Policies, Meta teams established a formal pathway for input from Trusted Partners within the slur designation process, with these civil society experts submitting approximately 60 terms for review across four regions.
We also invited the Board's guidance at an earlier stage of policy and product development than ever before. As we prepared to expand Community Notes internationally, we requested a Policy Advisory Opinion on what factors to consider when deciding whether a country should be included in the rollout, incorporating external expertise before deployment rather than after. The Board's input is supporting how we sequence that expansion.
In May 2026, Meta committed additional funding to sustain the Board's work through 2028. As the Board’s impact and influence progresses, it is no longer only an accountability mechanism for Meta; it has also become a governance resource whose expertise informs how the technology sector as a whole approaches the hardest questions it faces.
To read the full Meta H1 2026 Bi-Annual Report and Appendix on the Oversight Board, click here: