Meta has published its Community Standards Enforcement Report for the first half of 2026, disclosing statistics on the moderation of user content and accounts on Facebook and Instagram: the volume of removals, enforcement accuracy, the number of incorrect actions, appeals and restorations, as well as data on specific violation categories.

According to the report, Facebook and Instagram removed 700 million pieces of content, while Facebook took action against 2.02 billion fake accounts, 12% more than in the second half of 2025 and 20% more than in the first half of 2025. At the same time, removals declined in several categories. For example, Facebook spam fell by 69% year over year.

Meta restored content less frequently after appeals in some categories. For adult content, the share of content restored after appeals fell over the year from 64.7% to 17.3% on Facebook and from 58% to 14.1% on Instagram.

Spam showed a different trend: 8.12 million pieces of spam content were restored during the period, compared with 2.57 million appeals. Some restorations occurred without appeals, so these figures cannot be directly treated as the share of successful appeals.

Overall moderation metrics

Meta uses aggregate metrics to assess moderation performance across multiple areas at once. The company analyzes Facebook and Instagram data to determine the scale of content removals and the effectiveness of enforcement against different violations.

In the first half of 2026, Facebook removed 487 million pieces of content and Instagram removed 213 million. By comparison, the figures for the second half of 2025 were 530 million and 223.8 million, respectively.

As a result, the volume of removed content in the first half of 2026 declined compared with the second half of 2025, by approximately 8.1% on Facebook and 4.8% on Instagram.

How accurately Meta enforces its rules

In addition to the volume of removed content, Meta assesses moderation accuracy separately for Facebook and Instagram. This metric shows how often the platform correctly identifies violations and takes appropriate action.

Overall, moderation accuracy across both platforms remained around 90% or higher during most periods. At the same time, Facebook and Instagram metrics differed noticeably. For example, in the first quarter of 2026, Facebook’s moderation accuracy fell to 85.19-87.02%, while Instagram reported 91.19-92.49%.

Meta tracks the number of false positives, cases in which the system mistakenly identifies content as violating the rules and takes action against it. In the first half of 2026, Facebook had approximately 57.9-59.2 million such cases, while Instagram had 18.5-19.3 million. By comparison, in the second half of 2025, the figure was 42.4-62.1 million on Facebook and 29.9-39.1 million on Instagram. Therefore, the volume of false positives on Facebook remained at roughly a comparable level, while it declined noticeably on Instagram.

Adult content

Meta restricts the publication of sexually explicit content on Facebook and Instagram. In the first half of 2026, Meta took action against 44.4 million pieces of content on Facebook and 24.2 million on Instagram.

As a result, the number of pieces of content against which Meta took action rose noticeably in the first half of 2026. The figure increased by approximately 34% on Facebook compared with the second half of 2025 and by nearly 67% on Instagram.

The most notable increase came in the second quarter of 2026. From April through June, Meta took action against 27.8 million pieces of content on Facebook and 16.1 million on Instagram. By comparison, from January through March, these figures were 16.6 million and 8.1 million, respectively.

Appeals and restoration of adult content

Users can appeal Meta’s decisions to remove adult content. In the first half of 2026, approximately 1.32 million appeals were filed on Facebook and 1.89 million on Instagram. Meta restored 229K pieces of content on Facebook and 266K on Instagram.

Over the year, the number of appeals declined noticeably. On Facebook, it fell from 4.9 million to 1.32 million, while on Instagram it decreased from 3 million to 1.89 million. At the same time, the volume of restored content dropped sharply, from 3.17 million to 229K on Facebook and from 1.74 million to 266K on Instagram.

The share of restored content changed particularly significantly. While it was 64.7% on Facebook and 58% on Instagram in the first half of 2025, a year later it had declined to 17.3% and 14.1%, respectively. Thus, the metric decreased by nearly fourfold on Facebook and by more than fourfold on Instagram. This indicates that Meta restored content much less frequently in 2026 after reviewing removal decisions.

Fake accounts

Meta aims to remove as many fake accounts as possible on Facebook. These include accounts created to violate the rules or cause harm, as well as profiles that impersonate a business, organization, or nonexistent person as a real user. The company pays particular attention to accounts linked to spam and financial fraud.

In the first half of 2026, Meta took action against 2.02 billion fake accounts on Facebook. By comparison, the figure was 1.8 billion in the second half of 2025.

At the same time, Meta detects nearly all fake accounts on its own. In the latest available period, October through December 2025, the company independently identified 99.9% of violating accounts, while user reports accounted for only 0.1%.

Keep in mind that these statistics cover accounts that had already been created and against which Meta took action. The company also blocks millions of attempts to create fake accounts every day, but those cases are not included in the figures above.

In its Adversarial Threat Report for the first half of 2026, Meta also described an unusual case involving the mass creation of fake accounts. In Pakistan, the company discovered approximately 575K mostly inactive accounts that it believes may have been created in advance and retained as a “reserve” for later use in fraudulent schemes. Meta notes that it will continue monitoring such practices because they may point to a new model for preparing accounts for large-scale fraud operations.

Spam

Meta defines spam as content distributed in a deceptive or intrusive manner, or used to artificially increase engagement. Spam can be distributed automatically through bots and scripts, or when multiple accounts are used to promote the same content. According to Meta, the main goal of such schemes is to increase audience size and reach, usually for financial gain.

At the same time, Meta cannot yet reliably assess the prevalence of spam on Facebook. The company explains that standard reviews of individual posts do not make it possible to fully identify violations based on account behavior and coordination between accounts.

In the first half of 2026, Meta took action against 162.9 million pieces of spam content on Facebook and 97.7 million on Instagram.

On Facebook, the volume of removed spam content continued to decline sharply: compared with the first half of 2025, the figure fell by approximately 69%, and by 44% compared with the second half of 2025. On Instagram, the year-over-year decline was approximately 47%, while the volume remained virtually at the same level compared with H2 2025.

For spam content in H1 2026, approximately 1.68 million appeals were filed on Facebook, after which a total of 6.33 million pieces of content were restored. On Instagram, there were 892K appeals and 1.79 million restorations. A significant share of restorations occurred without an appeal.

• Full report: https://transparency.meta.com/reports/community-standards-enforcement/