Industries / Technology / Social media
Safety for young users
This work decides who is old enough to join a platform, what children's and teenagers' accounts can see and do by default, and who answers for harm when those protections fail.
7
Milestones achieved, from 2008 to 2026.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
2008
Peers found the bigger risk
Under a January 2008 agreement between MySpace and the US state attorneys general, a task force directed by Harvard's Berkman Center reviewed the research on young people online and 40 technologies that companies submitted. Its final report, dated 31 December 2008, found that bullying and harassment, "most often by peers," were the most frequent threats minors faced, and its technology board judged age verification "appealing in concept but challenged in terms of effectiveness."
LessonChecking ages was judged no answer on its own, a question that returned when laws began to require age checks in 2025.
Credit
Internet Safety Technical Task Force, chaired by John Palfrey, with danah boyd leading its research board
2021
A design code for children
The UK Information Commissioner's Office issued its Age Appropriate Design Code, also called the Children's Code, on 12 August 2020 and gave online services until 2 September 2021 to comply. It sets 15 standards for services likely to be used by under-18s, among them settings that are "high privacy" by default, location services switched off by default, and no nudges that push children to weaken their privacy settings.
Credit
UK Information Commissioner's Office
2022
A coroner blames online content
On 30 September 2022 the inquest into the death of Molly Russell, a 14-year-old from north London who died in November 2017, concluded that she "died from an act of self-harm whilst suffering from depression and the negative effects of on-line content." The coroner found that platforms' algorithms had at times given her "binge periods" of images, video clips and text about self-harm and suicide, some of which she had not asked for, and sent a report on preventing future deaths to the UK government, Meta, Pinterest, Snap and Twitter.
LessonThe harm lay partly in how much content was pushed and how fast, not only in single posts.
Credit
Andrew Walker, senior coroner for North London
2025
Child safety duties begin
The UK's Online Safety Act became law on 26 October 2023. After its codes for protecting children were laid in Parliament on 24 April 2025, services likely to be used by children had until 24 July 2025 to assess the risks to them, and the Act requires social media companies to enforce their own minimum ages consistently. The government's explainer says harm can come from the way content is spread, "such as when an algorithm repeatedly pushes content to a child in large volumes over a short space of time," and that fines can reach £18 million or 10% of qualifying worldwide revenue, whichever is greater.
Credit
UK Parliament; Ofcom, the regulator that writes the codes
2025
A minimum age of 16
Australia's Online Safety Amendment (Social Media Minimum Age) Act 2024 requires age-restricted platforms to take "reasonable steps" to stop under-16s holding accounts, from 10 December 2025. Citing preliminary figures from the eSafety Commissioner, the government said more than 4.7 million under-16 accounts were deactivated, removed or restricted within days. In March 2026 eSafety's first compliance report found that a "substantial number" of children still held accounts or passed platforms' age checks, named poor practices such as allowing unlimited attempts at an age check, and began gathering evidence against Facebook, Instagram, Snapchat, TikTok and YouTube.
LessonAn age check that can be retried without limit will in time be passed.
Credit
Parliament of Australia; eSafety Commissioner Julie Inman Grant
2026
Juries fault platform design
On 24 March 2026 a jury in Santa Fe found that Meta had broken New Mexico's Unfair Practices Act by misleading the public about the safety of its platforms for children, and set civil penalties of $375 million, the $5,000 maximum for each of thousands of violations. The next day a Los Angeles jury, in one of three bellwether trials chosen to show how thousands of similar claims may go, found Meta and YouTube negligent in the design of their platforms and awarded $3 million in compensatory and $3 million in punitive damages, split about 70% and 30%. Both companies said they would appeal.
LessonBoth cases turned on product design and the companies' own statements, not on users' posts, for which Section 230 of US law has long shielded platforms.
Credit
New Mexico Attorney General Raúl Torrez, who brought the state's case in 2023; the two juries
2026
States settle with Meta
On 26 August 2026 a federal court approved a settlement that, by New Jersey's count, resolves the claims of 47 states, the District of Columbia and three territories against Meta, ending a trial that had opened in Oakland, California, the week before. New Jersey's attorney general puts the payments at $12.1 billion to $17.1 billion; Meta's own count is up to about $18.59 billion across its agreement with 52 attorneys general, paid over 10 years. Meta must cap young users' combined Instagram and Facebook time at two hours a day for five years, block their access from midnight to 6 a.m., limit notifications in school hours, restrict beauty filters and visible "like" counts, and accept independent audits; the cap falls to 60 minutes if Snapchat, TikTok and YouTube adopt comparable terms.
Credit
the attorneys general of California, Colorado, Kentucky and New Jersey, who led the trial team
Watch for
Teen protections, independently checked
Meta says teens placed in Instagram's Teen Accounts "saw less sensitive content, experienced less unwanted contact," and on 25 September 2025 it extended the accounts to users worldwide. The same day a report by Arturo Bejar, a former Meta employee, with Cybersecurity for Democracy at New York University and Northeastern University, the Molly Rose Foundation, Fairplay and ParentsSOS, said that of 47 of Meta's 53 listed teen safety tools it tested, only eight worked as intended with no limitations. Meta said the report "repeatedly misrepresents our efforts."
Achieved whenthe independent audits required by the August 2026 settlement publish measured results for the time limits, night blocks and content safeguards.
Claimed by
Meta, September 2025
Australia weighs court action
In March 2026 the eSafety Commissioner, Julie Inman Grant, said her office had "significant concerns about the compliance" of five of the ten platforms covered and was gathering evidence that they had not taken "reasonable steps" to keep under-16s off them. A court can order fines of up to A$49.5 million for systemic failures to comply.
Achieved whena court rules on whether a platform's steps were reasonable.
Claimed by
eSafety Commissioner, March 2026
Open
01
Track teen safety promises
A dated public archive of the pages where ten large platforms (Instagram, Facebook, TikTok, YouTube, Snapchat, X, Reddit, Discord, Pinterest and Roblox) describe their teen accounts, default settings and parent controls, captured weekly as they appear in the UK, the EU, Australia and the US, with each change classed as a tool added, a tool removed or a default changed. Meta's August 2026 consent judgment with the states serves as the first checklist: for each obligation, among them the two-hour daily cap, the pauses after 15, 60 and 90 minutes of use, the midnight-to-6 a.m. block, the school-hours limit on notifications, hidden "like" counts and limits on beauty filters, the archive records the date Meta's help pages and release notes first describe it, and whether Snapchat, TikTok and YouTube adopt the comparable terms that would tighten the limits and, by Meta's account to AP, release about 30% of the payment.
The briefCredit · unclaimed02
Trace age checks in apps
A public record of what leaves an adult tester's own phone when a social app asks for an age check, covering the checks that adults in the UK and Australia meet on at least eight platforms. For each check the team captures the network traffic on a test phone it owns and records which companies' servers receive data, whether a face image or ID scan is uploaded or only a pass or fail result, and how that compares with what the app and the age-check company say. Only the testers' own faces and documents are used.
The briefCredit · unclaimed
Open 01 · the brief
Track teen safety promises
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A dated public archive of the pages where ten large platforms (Instagram, Facebook, TikTok, YouTube, Snapchat, X, Reddit, Discord, Pinterest and Roblox) describe their teen accounts, default settings and parent controls, captured weekly as they appear in the UK, the EU, Australia and the US, with each change classed as a tool added, a tool removed or a default changed. Meta's August 2026 consent judgment with the states serves as the first checklist: for each obligation, among them the two-hour daily cap, the pauses after 15, 60 and 90 minutes of use, the midnight-to-6 a.m. block, the school-hours limit on notifications, hidden "like" counts and limits on beauty filters, the archive records the date Meta's help pages and release notes first describe it, and whether Snapchat, TikTok and YouTube adopt the comparable terms that would tighten the limits and, by Meta's account to AP, release about 30% of the payment.
Why it is still open
The settlement's independent audits are "reported to and reviewed by concerned states," with no promise of publication. Open Terms Archive already keeps dated copies of platforms' terms of service and EU transparency reports, but none of the 44 document types in its shared list covers teen safety or parent controls, and the September 2025 test of Instagram's tools had to start from Meta's list of 53 tools as it stood at one moment. A GitHub search found no tracker for the settlement.
Built with what exists
Open Terms Archive's open-source engine with a new collection, or changedetection.io; a browser agent such as Browser Use 1.0 for pages that load by script (see the Note "Deleting 80% to start over"); a language model such as Claude Fable 5.2 (see the Note "Three working devices") to match changed text to each term, with a person confirming every match; a VPN exit in each region.
First step
Copy the judgment's obligations into a spreadsheet, one row per term with its deadline, and declare Meta's teen help pages in an Open Terms Archive collection for a first capture.
Done looks like
Ten platforms in four regions captured weekly for six months, every term of the judgment marked not described, described or described as live with a dated, archived link as evidence, and a public change log in which a second person has checked the class given to each change.
Open 02 · the brief
Trace age checks in apps
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public record of what leaves an adult tester's own phone when a social app asks for an age check, covering the checks that adults in the UK and Australia meet on at least eight platforms. For each check the team captures the network traffic on a test phone it owns and records which companies' servers receive data, whether a face image or ID scan is uploaded or only a pass or fail result, and how that compares with what the app and the age-check company say. Only the testers' own faces and documents are used.
Why it is still open
"Papers, Please", a study by Shreyas Minocha, Isaac Sheridan and three professors at Georgia Tech and UC Irvine, presented at the IEEE Symposium on Security and Privacy in May 2026, traced age checks on websites covered by US state laws. It estimated that one vendor, Yoti, served about 60% of the sites that used an outside service, and reported that its checks could send IP addresses and device details to other companies. Yoti's chief executive, Robin Tombs, called claims that it passes face images to third parties "wholly false" and invited an independent audit. The study looked at the web, not at the apps where most social media use happens.
Built with what exists
A rooted Android test phone the team owns, running mitmproxy or HTTP Toolkit to record its own traffic; the apps' own age-check flows; a VPN exit in each country; DuckDuckGo's open Tracker Radar list to name the companies behind each server.
First step
Set up mitmproxy on a rooted Android test phone and capture one age check that a UK adult account meets when it opens content marked for adults.
Done looks like
Checks on at least eight platforms in two countries, each run three times by two adult testers, with a published table of every receiving domain, the company behind it and the kind of data sent, plus the capture method and the companies' replies.