Digital Guardians Safety System

The Safety Settings Were Tested. Here's What Actually Held Up.

Researchers tested 86 teen-safety features across Instagram, Snapchat, TikTok and YouTube. Fifty-one failed — but the thirty-five that worked share a…

The Safety Settings Were Tested. Here's What Actually Held Up.

Every major platform your child uses publishes a list of the tools that keep them safe. Teen accounts. Content filters. Screen-time limits. Message restrictions. It is a reassuring list, and most of us have skimmed it, flipped a few switches, and felt a little better.

In June 2026, a team of researchers did something almost nobody had done before: they took those lists and checked them. Not the marketing — the actual features, one at a time, on real phones, using accounts registered as minors. They asked two questions of every tool. Does it do what the company says it does? And could a child realistically find it and use it?

The answer is worth a parent's attention, and it is not the answer you might brace for. It is not "none of this works." It is something far more useful: some of it works very well, and the ones that work share a pattern you can act on tonight.

Someone Finally Tested the Lists

The study, Broken, Buried, or Missing, came from the Cybersafety Research Center — a joint effort of NYU's Tandon School of Engineering and Northeastern's Khoury College — working with former Meta engineering leader Arturo Béjar. They tested 86 safety features across Instagram, Snapchat, TikTok and YouTube between December 2025 and June 2026. Fifty-one failed. Thirty-five worked.

The failures sorted into three plain categories. Nine features were missing — researchers followed the company's own instructions and found no evidence the tool existed. Thirty-four were broken — the feature was there, but failed in a way that defeated its purpose or could be sidestepped with trivial, non-technical effort. Eight were buried — they worked, but were nested so deep that expecting a child to use them was unreasonable.

A year earlier, a separate team had done the same exercise on Instagram alone, testing 47 of Meta's own listed features and publishing through the Molly Rose Foundation, Fairplay and ParentsSOS in September 2025. Their result was similar in shape: eight features worked as advertised, nine worked with serious limitations, and thirty were rated as gone or easily circumvented.

Two findings from the 2026 study deserve to be said plainly, because they change what a parent should expect. First, every one of the ten tools designed to govern how users treat each other — the anti-bullying safeguards — failed on all four platforms. Second, tools meant to curb compulsive use fared little better: fewer than one in three succeeded, because most of them let a teen dismiss the limit and keep scrolling.

What the Companies Say

Meta disputes both reports, and that belongs here. Responding to the 2026 study, the company said the authors "include vague claims that our features are broken but, in the vast majority of cases, either misrepresent those features or fail to provide any examples or evidence," and pointed to Teen Accounts, saying teens "are seeing less sensitive content, experiencing less unwanted contact, and spending less time on Instagram at night." On the 2025 report, Meta called it "misleading" and "dangerously speculative." Snap, TikTok and Google also contested the 2026 findings.

We print that because you deserve both halves. Digital Guardians takes no funding from any technology platform, which is exactly why we can publish a company's rebuttal without flinching — and publish the research without flinching either.

The Pattern in What Worked

Here is the finding that actually helps you. Thirty-five features did work, and the researchers found they mostly shared two design traits.

The first is defaulting to the safest path. Protections that apply automatically, before anyone chooses anything, cannot be missed, forgotten or buried. Instagram setting teen accounts to private the moment they are created is named in the report as an example that works precisely because nobody has to find it.

The second is removing a risk rather than filtering it. Almost every documented failure was a containment failure: a filter that missed a spelling, a search block that got recommended around, a friend suggestion that slipped through. The report contrasts this with reduced products like TikTok for Younger Users, where open search, messaging and the algorithmic feed simply do not exist. A restriction cannot be circumvented if the capability it guards was never offered at all.

For Parents: When a platform offers both a stripped-down account for younger users and the full app with protections layered on top, choose the stripped-down account — even though it looks like the lesser product. Its safety does not depend on a filter working perfectly every single time. That one choice outperforms a dozen toggles.

What Changes in the Next Year

On August 26, 2026, Meta reached an agreement with a bipartisan coalition of state attorneys general, entered as a court order. For families it matters for one reason: it converts several protections from things you have to hunt for into defaults that arrive already switched on — a two-hour daily limit counted across Facebook and Instagram together, a block on access between midnight and 6 a.m., muted notifications during school hours, and like counts hidden by default.

Two caveats keep this honest. The Connecticut Attorney General's office says these changes must arrive "over the next year, and in some instances much sooner" — meaning they are commitments, not yet the experience on the phone in your child's hand. Set those limits yourself in the meantime. And they apply to Meta's apps only. Meta has publicly called on TikTok and YouTube to match the framework; as of early September 2026, no equivalent commitment from TikTok, YouTube or Snapchat has been announced.

Four Moves That Work Anywhere

Prefer the settings that are already on, and confirm they really are. Prefer removing a capability over filtering it. Assume the filter will miss something, because testers got past keyword blocks in under three minutes on more than one platform. And check it yourself, once, on your child's device — open their account, search for something the platform promises to block, and see what comes back.

For Kids: "I read that some of the safety stuff these apps advertise doesn't work as well as they claim — researchers actually tested it. So I'm turning on the parts that do work. This isn't because you did anything wrong. And if something weird ever gets through, tell me. You will not be in trouble for telling me. Ever."

The Setting Is Not the Safety Net

Every finding in this research describes something getting past a filter. What determines how badly that lands is almost never the setting. It is whether your child tells someone, and how fast.

That is the part no company can build for you, and it is the part that never needs a software update. Settings are a layer. The relationship is the net.

We keep a full, dated breakdown — every platform, every claim, every source, and what would make it out of date — on our Platform Watch page. It is reviewed every quarter, and it tells you exactly when it was last checked.

---

📖 Part of the Digital Guardians Safety System — Pillar 3: Device, App & Network Protection.

Take the next step. Know what your settings do before you rely on them.