Meta cross-check: the enforcement-exemption tier

Start with the volume, because everything in this case follows from it. Meta told the Oversight Board it was performing about 100 million enforcement attempts on content every day at the time of the Board's briefings. The Board did the arithmetic in the opinion: at 99 per cent accuracy, that is a million mistakes a day. Four years later Meta put a number on the real rate from the other end, saying in January 2025 that of content actioned in December 2024, one to two out of every ten actions may have been mistakes.

Now add a tier. When Meta's systems identify content as breaching a Community Standard, an ordinary account's post is actioned. A listed account's post is not. It stays fully accessible while additional humans reconsider it. Meta calls the entity-list pathway Early Response Secondary Review — renamed Secondary Sensitive Entity Review effective 25 April 2024 — and its content can pass through as many as five layers before any enforcement: initial automated or at-scale human identification; the Regional Market Team, staff and contractors with language and market knowledge; the Early Response Team, the first layer that may authorise enforcement, which holds the escalation-only policies and the newsworthiness and spirit-of-policy allowances but is not required to have language or regional expertise and works from Market Team notes and translations; an enhanced Early Response Team review with content-policy subject-matter experts plus Public Policy, Communications and Legal; and a discretionary global senior-leadership review on liability, urgency, geopolitical impact or service-blocking risk. The process stops and the content stays up the moment any reviewer finds it non-violating.

Everyone else's flagged content goes to General Secondary Review, built in 2021 and extended to content from all users in early 2022. An automated ranker scores the item on how sensitive its topic is, how severe the enforcement action is, how likely that action is to be a false positive, its predicted reach and how sensitive the entity behind it is — with the topic score and the entity score weighted most heavily — and either suspends enforcement and queues the item or lets the original action stand. Approximately 0.01 per cent of all content identified as needing enforcement was escalated through cross-check to reviewers empowered to apply context-specific policies and allowances. Meta also told the Board that 10 per cent of organic content otherwise subject to integrity enforcement is not eligible for cross-check review at all.

Here is the engineering fact the whole case turns on, and Meta disclosed it: both pathways draw on the same Regional Market Team capacity, the entity-list pathway is guaranteed and served first, and the same teams carry non-cross-check work besides. So a single capacity shortfall shows up twice, in opposite forms.

On the listed pathway it shows up as time. Meta's internal goal for Market Team cross-check review was 12 to 120 hours by severity. The actual mean time to a final Early Response Secondary Review decision was more than five days; approximately 12 days on average for content posted from the United States; about 17 days on average for Afghanistan and Syria; and the longest single piece of content in the queue in the data Meta supplied had sat 222 days. Meta's own position is that most views happen when content is fresh. So the injury on this pathway is not a wrong verdict. It is the interval before a right one, spent in the window where the views are.

On the general pathway the same shortfall shows up as its mirror image. Content the reviewers do not reach within two to four days times out, and the original enforcement action is applied with no human having looked. That pathway overturned about 80 per cent of what it reviewed at the February 2022 briefings and about 70 per cent for March to May 2022, from which the Board concluded that when General Secondary Review content times out "it is highly likely that Meta is enforcing a significant number of false positives". Meta told the Board it "never intended to operate with a consistent backlog" but that "operational capacity constraints and increasing volumes have led to a backlog in Early Response Secondary Review". Its own research identified the drivers of users seeing violating content as incorrect overturns and the delay of enforcement for non-overturns slowed by the secondary review process. Read those two pathways together and the shape is plain: protection for a few, paid for in mistaken removals against everyone else, with no ledger anywhere connecting the two.

Then ask who is on the list. The opinion records that entitlement "is also determined by an assessment of the impact a potential enforcement mistake would have on the company in terms of the level of company leadership that would be involved in finding a solution", with "escalation at the highest levels (CEO, COO)" corresponding to an "extremely high severity" tag; that inclusion criteria referenced advertising spend or revenue across the family of apps, enforcement history and follower reach; and that a single Meta employee could add an entity to a cross-check list with no required review of that decision. Entities carry tags in six categories — civic and government; significant world events; media organisations, businesses, communities and creators including advertisers; historically over-enforced; legal and regulatory; and entities under review — which generally expire after a year, with civic-and-government tags carrying no default expiry. The people on the list are not told they are enrolled. And the allocation followed the criteria: 42 per cent of content reviewed on the entity-list pathway came from the United States or Canada, and 20 per cent of listed entities corresponded to those two countries, against 9 per cent of Facebook monthly active people.

Beside cross-check sits a harder exemption Meta calls technical corrections: automatic bars that override almost all enforcement attempts for a preselected entity-and-violation pair, checked before cross-check eligibility is even assessed. Meta applied about a thousand a day and said four remained active as of 21 September 2022, primarily on spam or inauthentic behaviour and impersonation, and acknowledged that a past "lack of governance over practices [...] inadvertently resulted in some entities not receiving many enforcement actions". Journalism had described this as allowlisting or whitelisting; the Board agreed the label is apt.

The Board learned that any of this existed only because it asked a question during the Trump case, which Meta had not disclosed in its referral. After The Wall Street Journal's September 2021 reporting from Frances Haugen's document set — which described the programme as exempting influential users from normal moderation, reported that internal staff considered the whitelisting practice "not publicly defensible", and reported a 24-hour self-remediation window — the Board found that "the team within Facebook tasked to provide information has not been fully forthcoming in its responses on cross-check". Meta requested a policy advisory opinion days later. That reporting is journalism recounted inside the opinion, which does not contest it, and it is used here at that weight and no more; the widely circulated figure for how many users the programme covered rests on secondary coverage, is not in the opinion, and is not asserted anywhere in this atlas.

What the Board could obtain, and what it could not, is the sharpest thing in the record. Meta refused its repeated requests for the entity list, citing user-privacy obligations; almost five months later it supplied only aggregate fields — entity type, self-selected country and language, a civic flag, a partner flag — and for a quarter of the listed Instagram entities disclosed only that they existed. Meta answered 58 of the Board's 74 questions fully, 11 partially and 5 not at all. And the Board's own reach into the programme was itself tiered: for May and June 2022 an average 35 per cent of cross-check content could not be escalated to it at all, so the accounts with the most reach had the content least able to travel outward.

The opinion of 6 December 2022 found four shortcomings — unequal treatment of users, delayed removal of violating content, failure to track core metrics, and lack of transparency — and concluded that while Meta told the Board cross-check advances its human rights commitments, "the program appears more directly structured to satisfy business concerns". On the rulebook it wrote: "Meta has repeatedly told the Board and the public that the same set of policies apply to all users. Such statements and the public-facing content policies are misleading." It issued 32 recommendations, the largest set it had issued at once. Two findings define the governance shape more than any other. Meta "did not provide the Board with information showing that it tracks data about the accuracy of decisions made through its cross-check system", and had no statistically significant data distinguishing account-level penalties applied to cross-checked versus non-cross-checked entities — so the programme's founding claim went untested by its own operator, and its single efficacy metric was an overturn rate on the content it routed, which ranged from 30 to 90 per cent on the entity pathway across periods in the year before the opinion. And on the remedy the Board itself obtained: internal auditing without external oversight falls short. There is no external audit of cross-check anywhere in this record.

What the review moved is real, and it is recorded independently by the Board's own Q2 2023 transparency report rather than by Meta alone. Meta "has cleared all outstanding backlogs in its cross-check review queues dedicated to potentially violating content from entities on its lists", "producing a 96% decrease in resolution time (time taken for review and any subsequent enforcement) for 90% of the jobs created in the first half of 2023, compared with the second half of 2022"; and the new technical-corrections approach "led to an immediate decrease in the overall size of the technical corrections list by more than half (55%)". List governance gained add-and-remove criteria, time-bound tags, multi-person approval and internal audit, replacing the state in which one employee could act alone. Meta committed to staffing cross-check decisions with reviewers who speak the language and have regional expertise.

What the review could not move is equally specific, and it is a list of five. Meta's own tracker records recommendations 5, 6, 12, 13 and 29 as "No Further Action": an open, criteria-based application route into the programme; an explicit rules re-commitment at enrolment; publicly marking the accounts of state actors, political candidates, business partners, media actors and commercially included public figures; telling a user who reports such an account's content that special procedures and longer timelines apply; and publishing metrics quantifying the adverse effects of delayed enforcement, such as views accrued on content left up during enhanced review and later found violating. Meta cited targeting and gamification risk for the two marking-and-notice items, and pointed to a promised cross-check-specific annual report under recommendation 30 in place of the harm metric. Read the five together and they are not a random subset: they are exactly the recommendations that would have made the tier visible to the people on the other side of it, or priced it.

There is a count to state and not resolve. The Board's opinion and its annex enumerate 32 recommendations, and Meta's own tracker page also enumerates 32; Meta's Q1 2023 quarterly update says it "responded publicly to all 33 of the board's cross-check recommendations, committing to implementing 82% either in part or in full". Both counts stand here as their authors wrote them. Aggregate implementation tallies for this opinion are unstable across renderings of Meta's tracker and across contemporaneous press summaries, so none is asserted; only the per-recommendation statuses that reproduced consistently and are corroborated by the downloaded quarterly-update PDF are used.

The promised report has not been located. Meta's cross-check recommendation tracker was last updated 3 October 2024. Its H2 2025 bi-annual report on the Oversight Board, published 19 March 2026 and covering 326 recommendations responded to as of 31 December 2025, contains no cross-check reporting. No annual report on the programme's functionality and impact was located as published as of 28 August 2026 — which is a statement about a search, not about Meta's intentions. Meanwhile the programme grew: in March 2025 Meta's cross-check teams sought the Board's input on expanding coverage to more users, with the result including further investment in a Dynamic Multi-Review system intended to reduce over-enforcement at scale while keeping sensitive activism and journalism content with specialised reviewers.

A population-level over-enforcement metric did arrive, but not the exemption-path one. Meta began publishing global enforcement precision in 2025, reporting around 91 per cent on Facebook and around 92 per cent on Instagram at the end of H1 2026, and reported roughly a 50 per cent reduction in United States enforcement mistakes between Q4 2024 and Q1 2025 after its 7 January 2025 policy overhaul — the one in which it ended third-party fact-checking in the United States and narrowed automated enforcement to illegal and high-severity violations. None of it is broken out for cross-check.

No court and no regulator has adjudicated the programme. The nearest live regulatory matter is adjacent to it rather than about it: on 24 October 2025 the European Commission issued preliminary findings, conducted with Coimisiun na Mean, that Facebook and Instagram appear not to provide a user-friendly, easily accessible notice-and-action mechanism for illegal content and appear to use dark patterns in it, and that their appeal mechanisms appear not to allow users to provide explanations or supporting evidence, with exposure up to 6 per cent of total worldwide annual turnover if confirmed. Those findings expressly do not prejudge the outcome, and they do not concern cross-check. What is worth recording is a coincidence of channels rather than a legal connection: the reporting channel and the appeal channel the Commission is examining are the same two channels the Board found this tier quietly bypasses.