Meta published its first detailed enforcement tally under Australia's under-16 social media law on August 12, 2026, disclosing that access has been removed from 756,000 Facebook and Instagram accounts between December 1, 2025 and June 30, 2026. The company used the same post to argue that age checks belong at the device or app store level rather than inside individual apps.

Meta set out the numbers in a post on its APAC newsroom carrying the date August 12, 2026. According to Meta, the company has removed access to more than 750,000 accounts on Facebook and Instagram in Australia that it assessed as belonging to people below the minimum age of 16. The accompanying table is more precise than the headline figure: 462,000 Instagram accounts and 294,000 Facebook accounts, for a total of 756,000. The period covered runs from December 1, 2025 to June 30, 2026.

That window begins nine days before the law itself took effect. Australia's restrictions took force on Wednesday, December 10, 2025, covering nine named services and exposing platforms that fail to take reasonable steps to civil penalties of up to 150,000 penalty units, roughly 49.5 million Australian dollars. Meta began removing accounts before the deadline rather than on it, and the disclosure confirms how much of the total was front-loaded: according to the company, over 500,000 of the accounts were removed in the lead-up to the law taking effect, a figure it reported in January 2026. The remaining quarter of a million or so accrued across the following five months.

Enforcement is a rate, not a one-off purge

The shape of the data matters more than the headline. If roughly 500,000 accounts went in the first weeks and 756,000 by the end of June, then the monthly run rate after the initial sweep sits somewhere near 50,000 accounts. Meta states that enforcement is ongoing and that the numbers will continue to grow, and that it reports its enforcement to the eSafety Commissioner and will publish updates periodically.

Two absences are worth noting. The table covers Facebook and Instagram only. Threads, which sits inside the same age-restricted list under Australian rules and which requires an Instagram account to sign in, does not appear as a separate line. Meta also does not break the figures down by month, by age band within the under-16 cohort, or by the mechanism that triggered each removal, so the relative contribution of automated detection versus user reports cannot be read from the disclosure.

The measures behind those numbers

Meta describes age identification as a continuing process rather than a single check. According to the company, the figures reflect a series of improvements made through the year, building on an approach it set out in May 2026 and explained for families in June 2026. Five specific measures are listed.

AI-powered proactive detection scans profile content for contextual clues. "We use AI to analyse profiles - including posts, comments, bios and captions - for contextual clues that an account may belong to someone under 16, such as birthday celebrations or mentions of school grades," according to Meta. The examples given are behavioural and textual rather than biometric.

Simplified reporting covers the path by which parents and other users flag a suspected underage account, both inside the apps and through the Facebook and Instagram Help Centres.

AI-driven review of reports is the follow-on step. Meta states that reports of underage accounts are now supplemented by AI-driven review, which it says delivers higher accuracy and faster resolutions than human review alone. That is a claim about the substitution of automated adjudication for human adjudication in a workflow that ends in account removal, and Meta publishes no accuracy figures alongside it.

Preventing re-registration addresses the obvious circumvention route. According to Meta, its systems have been strengthened to detect and prevent someone from creating a new account after a previous account was removed.

App Store age rating is the only measure that sits outside Meta's own products: the company updated its Apple App Store age rating to 16 plus in Australia.

The technical direction is consistent with what Meta has been building elsewhere. In May 2026 the company extended AI age estimation using visual cues such as height and bone structure to Instagram in the European Union and Brazil and to Facebook in the United States, while stating that the system does not identify specific individuals and is therefore not facial recognition. The Australian measures described on August 12 rest on profile text and behavioural signals rather than on that visual layer, at least as described.

A campaign aimed at parents, not teenagers

Between June and July 2026, Meta ran an education campaign across Facebook and Instagram in Australia that it says reached approximately 1.3 million people. The campaign pointed to guidance on how underage accounts are detected and removed, what happens to a teenager's account and stored content, and how a stated age can be checked and corrected.

The company defends the choice of audience directly. "A young person using social media without a parent or guardian knowing is harder to support, not easier, and keeping parents informed is part of making these protections work," according to Meta.

Reach of 1.3 million in a country of roughly 27 million people is a substantial in-feed campaign, and it was delivered on the same inventory advertisers buy. Australian households with teenagers were, for two months, receiving compliance messaging in the same slots that carry commercial creative.

Meta wants the check moved to the device

The most consequential passage of the August 12 post is not about Australia at all. Meta states that its primary concern is that platforms are not taking a consistent approach to how they are removing teens, and argues for a different architecture: "There should be a single reliable age signal delivered at the operating system or app store level."

According to Meta, age could be verified once, when a young person sets up a device or downloads an app, rather than repeatedly across the dozens of apps they use.

That position is not new, and it is not uncontested. In June 2025, Google's global director of privacy, safety and security policy criticised the app store model, arguing that it would require sharing granular age band data with millions of developers who do not need it and that it is ineffective against pre-installed applications, a category into which Meta's own apps frequently fall. The two companies have a commercial as well as a philosophical stake in where the check sits. Moving age assurance to the operating system shifts cost and liability toward Apple and Google; leaving it inside apps keeps it with Meta, TikTok, Snapchat and every smaller service.

Australia's law, as written, places the obligation on the platform. The eSafety Commissioner has assessed 10 services as age-restricted platforms and a further 10 as outside scope, a split documented when the United Kingdom opened its own consultation on children's online experience in early 2026.

Where this sits in a widening regulatory map

Australia is the first jurisdiction with observable enforcement data, which is why its numbers travel. The United Kingdom announced on June 15, 2026 that it would bar platforms from serving children under 16, with a Spring 2027 target and the Australian statute as its explicit model. France approved an under-15 ban in July 2026 while removing the age verification clause that would have specified how the restriction is enforced, setting September 1 as its enforcement date.

Meta is meanwhile defending age-related findings in Brussels. The European Commission issued preliminary Digital Services Act findings on April 29, 2026 that Instagram and Facebook failed to keep under-13s off the services, and followed on July 10, 2026 with separate preliminary findings on addictive design naming infinite scroll, autoplay, push notifications and recommender systems. Confirmed non-compliance under the DSA carries exposure of up to 6% of global annual turnover.

Litigation in the United States has moved in parallel. A New Mexico court entered a $567 million abatement order over teen harm this month, and a federal judge ruled on August 5, 2026 that minors' feeds are not protected speech in a challenge to California's age-appropriate design law. Product changes have accumulated alongside the legal ones: Teen Accounts arrived in September 2024, restrictions tightened across Facebook and Messenger in April 2025, content filtering aligned with PG-13 movie standards in October 2025, and the 13 plus content tier went global in June 2026.

What 756,000 accounts mean for buyers

The removals land in a market that has been growing quickly. Australian digital advertising reached a record 4.9 billion dollars in the first quarter of 2026, up 15.3%, with social video holding 40% of total video investment in calendar 2025 at 2.2 billion dollars. Against that base, the direct revenue at stake in an under-16 cohort is small, because the cohort was not directly monetisable in the way older audiences are and because teen accounts already carried personalisation constraints.

The second-order effects are the ones media buyers will notice. Removing an age band changes the composition of engagement signals feeding recommender systems, alters household reach patterns in categories where teenagers influence purchases without making them, and reduces total time on platform in the market. Frequency distributions shift when a segment leaves. Lookalike seeds built on Australian audiences before December 2025 now describe a population that no longer exists on the platform in the same shape.

There is also a measurement question that the disclosure does not answer. Accounts removed by an age estimation system are removed on probability, not proof. Meta's own framing describes identifying age online as a continuing process rather than a single check, which is another way of saying that some proportion of the 756,000 will have been wrong in either direction. False positives generate appeals and reinstatements; false negatives leave under-16 users inside the audience an advertiser is buying. Neither error rate is published.

For the signalling layer, the direction of travel is consistent across vendors. Reddit disabled teen chat and ad personalisation in the EU as age checks began on June 24, 2026, and Google retired the TFCD and TFUA signals in favour of a TFAT signal with a dedicated teen tier in May 2026. Age is becoming a first-class field in ad serving rather than a compliance flag bolted onto campaign setup.

The unresolved part

Meta's post ends with an argument rather than a commitment. The company says it shares the Australian government's goal, that it is meeting its obligations, and that it will continue to invest in more reliable age understanding while protecting privacy. It also says platforms are not taking a consistent approach to removing teens, which is a statement about competitors as much as about itself.

Nothing in the disclosure indicates whether eSafety considers 756,000 sufficient. The regulator has been collecting removal data from platforms since December 2025 under its own notice regime, and its assessment of what reasonable steps require has not been published alongside Meta's numbers. The company's financial exposure elsewhere gives the question weight: Meta reported second-quarter 2026 advertising revenue of $59.36 billion on July 29, 2026, with net income down 8% to $15.8 billion after a $2.4 billion legal charge. Australian penalties are small next to that. The precedent value is not.

Timeline

Summary

Who: Meta Platforms, operator of Facebook and Instagram, reporting to Australia's eSafety Commissioner under the Online Safety Act minimum age provisions. Australian parents, teenagers under 16, and advertisers buying social inventory in the market are the affected parties.

What: Meta disclosed that access has been removed from 756,000 accounts assessed as belonging to Australians under 16, split between 462,000 on Instagram and 294,000 on Facebook. Over 500,000 of those were removed before the law took effect. The company also detailed five enforcement measures, reported an education campaign reaching approximately 1.3 million people, and argued that a single age signal should be delivered at the operating system or app store level.

When: The post was published on August 12, 2026. The enforcement figures cover December 1, 2025 to June 30, 2026. The education campaign ran between June and July 2026.

Where: Australia, across Facebook and Instagram. The policy argument about operating system level age signals applies globally.

Why: Australia's under-16 restrictions took effect on December 10, 2025, requiring age-restricted platforms to take reasonable steps to prevent under-16s from holding accounts or face civil penalties of up to 49.5 million Australian dollars. Meta reports its enforcement to the eSafety Commissioner and has committed to periodic public updates.