--- title: "Most Child-Safety Features on Social Media Don't Actually Work, Research Finds" description: "A widening body of research — including a landmark audit led by a former Meta engineer — has found that the safety tools social platforms advertise to parents routinely fail in practice, with nearly two-thirds of Instagram's teen protections rated broken or nonexistent." category: "U.S." category_url: https://herald.la/category/us author: "Lucía Fuentes" published: 2026-06-29T13:48:00.000Z updated: 2026-06-29T13:48:00.000Z canonical: https://herald.la/article/most-child-safety-features-on-social-media-don-t-actually-work-research-finds tags: ["social media", "child safety", "Instagram", "Meta", "tech regulation", "KOSA", "online safety"] --- # Most Child-Safety Features on Social Media Don't Actually Work, Research Finds A widening body of research — including a landmark audit led by a former Meta engineer — has found that the safety tools social platforms advertise to parents routinely fail in practice, with nearly two-thirds of Instagram's teen protections rated broken or nonexistent. For years, social media companies have assured parents and lawmakers that they take child safety seriously, rolling out teen accounts, content filters and screen-time tools. A growing stack of research says most of those safeguards don't hold up. ## Promises vs. reality In a systematic audit of Instagram's safety toolkit, a coalition that included the advocacy group [Fairplay](https://fairplayforkids.org/instagram-teen-accounts-fail-to-protect-children-safety-tools-testing-reveals/), the Molly Rose Foundation, and the NYU/Northeastern research project Cybersecurity for Democracy — led by Arturo Béjar, a former Facebook engineering director turned whistleblower — tested 47 tools on Instagram's Teen Accounts. Thirty were rated red, meaning ineffective or nonexistent; nine yellow; and just eight green. By that math, about two-thirds of the tools failed, and only 17 percent worked as Meta had described them. The failures were not subtle. Testers found Instagram's autocomplete surfacing search terms tied to suicide, self-harm and eating disorders; screen-time tools that did little; and an algorithm that recommended Reels featuring young children to teen accounts. Béjar's conclusion: most of the tools were ineffective, unmaintained, or quietly removed. ## A pattern across platforms The Instagram audit did not stand alone. In a February 2026 test of six major platforms, [Malwarebytes researchers](https://www.malwarebytes.com/press/2026/02/10/cracks-in-social-media-platforms-allow-kids-under-13-access-to-content-on-fraud-drugs-and-other-illicit-activity) — simulating underage users who had bypassed age gates — reached explicit content on one service, fraud and identity-theft tutorials on another, and violent or sexually suggestive material on a third. Protections, they found, tended to work only as long as a child stayed precisely within the boundaries designed for them; a curious child who stepped outside quickly hit content the platforms claim to block. A separate academic analysis of years of platform safety announcements concluded that the companies' public claims were often misleading — touting features as global when they were regional, or citing statistics with no baseline. ## What the companies say Meta has not directly refuted the coalition's findings; after the audit it announced plans for an MPAA-style content-rating system to help parents, and it maintains that its Teen Accounts represent real progress and that it is continually improving. Other platforms named in the research did not issue public responses to those specific studies. ## A legal and regulatory vise The research lands amid intensifying legal pressure. More than 40 state attorneys general have sued Meta over harms to children, and courts have allowed several state suits against TikTok to proceed. In Congress, the path is murky: a House committee reached a bipartisan deal in June 2026 on a version of the Kids Online Safety Act that [stripped out its "duty of care" standard](https://thehill.com/policy/technology/5938935-house-breakthrough-on-kids-online-safety-faces-long-odds-in-senate/) — the requirement that platforms actively prevent foreseeable harm — and Senate sponsors pronounced that version dead on arrival, with Sen. Richard Blumenthal calling KOSA without a duty of care a "blank check" for the companies. For the researchers who ran the tests, the audit numbers and the legislative standoff tell one story: the companies have had years to make their safety tools work, and largely haven't. ## Sources - [Instagram Teen Accounts fail to protect children, safety tools testing reveals](https://fairplayforkids.org/instagram-teen-accounts-fail-to-protect-children-safety-tools-testing-reveals/) - [Cracks in social media platforms allow kids under 13 access to harmful content](https://www.malwarebytes.com/press/2026/02/10/cracks-in-social-media-platforms-allow-kids-under-13-access-to-content-on-fraud-drugs-and-other-illicit-activity) - [House breakthrough on kids online safety faces long odds in Senate](https://thehill.com/policy/technology/5938935-house-breakthrough-on-kids-online-safety-faces-long-odds-in-senate/)