Fined, Not Fixed: Meta's Untouched Algorithm
Meta’s large pockets allow it to happily pay fines over child safety everywhere, admit to paid political boosting, employ a two-tier fact-checking system, all while its ranking algorithm stays unchanged.
New Mexico's First Judicial District Court, in a ruling by Judge Bryan Biedscheid, raised Meta's penalty for failing children on its platforms to $942 million in August 2026, layering a $567 million abatement fund on top of an earlier $375 million civil penalty. The jury had found that Meta concealed what it knew about risks to children's mental health and about child sexual exploitation on its platforms, while telling the public and regulators that its products were safe. Weeks later, press investigations surfaced more than 300 ads on Meta's platforms carrying AI-generated child sexual abuse material (CSAM), prompting Senator Mark Warner to press chief executive Mark Zuckerberg on the failure.
It compels Meta to build predictive models for identifying under-13 users, request age verification from suspected minors, treat unverified accounts as children's accounts by default, delete data collected from verified under-13 users, open a reporting channel for schools and child-safety organisations, and file compliance updates twice a year, within a two-year window. Every obligation concerns identification, disclosure and data handling, but not how the algorithm ranks or amplifies content once a user is inside the feed.
Fines imposed on Meta worldwide, by year — All government penalties, converted to US dollars — 2020: 0; 2021: 255; 2022: 760; 2023: 1800; 2024: 610; 2025: 250; 2026: 942 — Compiled from regulators' published decisions. 2020: none. 2021: $255m, Ireland. 2022: $760m, Ireland. 2023: $1,800m, Ireland. 2024: $610m, Ireland, Nigeria and South Korea. 2025: $250m, European Commission and Nigeria. 2026: $942m, New Mexico. Excludes private settlements.
Meta ended its third-party fact-checking programme in the United States in January 2025, replacing it with a crowdsourced Community Notes model modelled on X, and said the US rollout would be "tried and tested" for a year before any decision on other countries. Nearly two years on, that US-first sequencing shows that Meta's certified fact-checking partners, accredited through the International Fact-Checking Network, continue operating in India and other markets. Indian fact-checking organisations supported by a nonprofit called Meedan depend on Meta funding as their largest single revenue source. American users have their claims democratically checked by an algorithm-plus-crowd system, while content elsewhere continues to be routed through gatekept American IFCN-accredited outlets that Meta still funds.
Last month, senior Meta executives including global affairs chief Joel Kaplan met Indian government officials following the removal of a post from the handle of Indian Prime Minister Narendra Modi. At the meeting where they were photographed arriving and leaving looking highly stressed, Meta representatives apologised for lapses tied to CSAM, deepfakes and operational failures, and separately acknowledged that money had changed hands to boost certain categories of political content. Zuckerberg himself apologised over the post removal within the three days the government had stipulated for the company to do so.
The acknowledgment followed a summer of pressure. In late July 2026, the Parliamentary Standing Committee on Communications and Information Technology summoned Meta alongside Google, X and Snapchat over how the platforms had been used to mobilise protesters during unrest linked to one of many paper leaks that have been an unfortunate occurrence across states and decades in the country. The platforms were seen to have weaponised initial sentiment to mobilise protests after the law had already taken its course and the guilty arrested, and the exam system itself digitised and overhauled. The street pressure was reminiscent of Arab Spring revolutions and managed to successfully topple a minister from the elected government as show of force. Meta, already conceding to accepting payment to amplify political content, is seen as pricing political influence while functioning as a foreign player in a market of 950 million Indian internet users.
Australia has taken the stricter route, removing the audience rather than regulating the platform's conduct. Its social media ban for under-16s took effect on 10 December 2025, requiring platforms including Meta's to prevent minors from holding accounts at all. An eSafety Commissioner study published in August 2026 found the ban still falling short: more than 80% of under-16s who had accounts before the ban retained them or opened new ones within three months, and daily usage among that age group barely moved, from around 60% to 58%. The regulator attributed the gap chiefly to platforms not checking ages in practice, even as officials continued defending the policy as having shifted the national conversation about children and social media. However, Tech Transparency Project, a nonprofit group looking into Meta’s CSAM control found that the company continued to overlook AI-generated sexual abuse material on minors, and WIRED’s reporting showed photos of real children being manipulated by AI for Meta to run ads. 120 of these ads were found to be in Australia, and 80 in India. BBC’s reporting on Meta also showed ads promoting CSAM continue to run unencumbered in India, and that “Just Rights for Children has filed a petition in India's Supreme Court, citing the BBC's findings, and asked for directions to social media platforms for mandatory reporting and identification of those promoting child sexual abuse material.” Just Rights For Children is an NGO founded by Bhuwan Ribhu fighting against the proliferation of CSAM material in India.
Penalties and settlements paid by Meta since 2020, by jurisdiction — Converted to US dollars. Saffron: government fine. Navy: private settlement. — European Union / Ireland: $3.3bn; Texas: $1.4bn; New Mexico: $942m; Illinois: $650m; Nigeria: $220m; South Korea: $15m — Compiled from regulators' decisions and court filings. The European figure aggregates Irish Data Protection Commission and European Commission fines from 2021 to 2025. Texas and Illinois resolved biometric privacy claims by settlement, not penalty. Nigeria's $220m, upheld on appeal in April 2025, is the largest imposed on a major Western platform by any country outside the US or Europe; the bar covers that penalty only, not the separate $32.8m data-protection order settled out of court in November 2025.
The problem is large enough that fines are merely a band-aid and sometimes a license. The EU has threatened Meta with fines over its addictive designs, and its lack of safety and has fined the American Big Tech organisation for antitrust violations. None of this has prevented Meta from continuing to operate its addictive algorithm promoting child sexual abuse material alongside decisively political content. Requiring algorithmic audits alongside age-verification mandates, obliging platforms to show, through data shared with regulators rather than self-reported metrics, that recommendation systems do not disproportionately surface high-risk content to accounts flagged as likely minors, alongside political sensitivity audits by independent bodies may offer nations battling Big Tech addiction a minor reprieve. Legislatures examining paid political content could mandate real-time disclosure of any payment tied to a post's reach, published in a public ledger rather than surfaced only under summons, so amplification becomes visible before an election rather than admitted after one. Age-verification laws could shift compliance burden and liability onto platforms at the point of account creation, with fines scaled to the gap between claimed and measured compliance, rather than relying on surveys to catch the shortfall months later. Each of these proposed fixes would need enforcement capacity that most regulators, including India's and Australia's, do not yet have.