Kids Safe Search Engines: What They Really Are

By Bill Hartzer · Updated 19 August 2026

Almost none of them are search engines

The single most useful fact about "kids search engines" is structural: with very few exceptions they do not crawl the web, do not hold an index, and do not rank anything. They are filtered front ends over Google or Microsoft Bing — a search box, a child-friendly skin, and a filtering layer applied to somebody else's results.

That is not a scandal in itself. Building and maintaining a web-scale index costs an enormous amount, and a small team that wants to make search safer for children has no realistic route to owning one. But it changes what you are actually choosing when you pick one. You are not choosing a different view of the web. You are choosing whose filter sits between a child and Google's results, and how much you can find out about the people who wrote that filter.

Two mechanisms do nearly all the work behind these sites:

SafeSearch — a setting on the underlying engine that suppresses explicit sexual content. It is Google's or Microsoft's classifier doing the work, not the kids site's.
A programmable or custom search instance — Google sells a product, now called Programmable Search Engine and previously Custom Search Engine, that lets anyone build a search box restricted to a chosen set of sites or the whole web, with their own ordering rules on top. A large share of "kids search engines" are one of these with a cartoon mascot.

On top of those, an operator can add its own domain allow-lists and block-lists, a list of forbidden query terms, and a handful of hand-picked results promoted to the top. Those additions are the entire proprietary content of most of these products.

What filtering can and cannot do

Filtering works reasonably well against the thing it was built for — explicit imagery arriving in response to an obvious query. It works much less well against everything else, and the gaps are systematic rather than accidental.

A safe domain can host an unsafe page. Filtering approves and rejects at the level of domains and keywords. Beyond a small number of hand-checked entries, no one has read the pages the search returns. A site that is broadly appropriate can carry a page that is not, and no allow-list catches that.

Keyword block-lists are trivially defeatable. A blocked term stops being blocked when a child misspells it or inserts a space. This is not a flaw in one implementation; it is inherent to matching strings against a list. Any product whose safety claim rests on a block-list has a safety claim that fails against a curious ten-year-old with a keyboard.

Blocking a query is not the same as helping. This is the most serious design question in the category and it is rarely asked. When a child searches for something about self-harm, suicide, sexual health or their own identity, a general engine typically surfaces crisis resources and helplines. A filtered engine that treats those words as forbidden returns a cartoon and an instruction to try something else. The child is not protected by that; they are redirected to a worse source. Whether any given kids engine surfaces crisis support or blocks the query is a specific thing an adult can test in thirty seconds, and it is worth testing before deploying one.

Filtering says nothing about what the safe results say. A filter is a test for explicitness, not for accuracy, balance or manipulation. Material that passes every filter can still be wrong or slanted, and no amount of tuning the block-list addresses that. This is the failure mode that content filtering cannot even in principle catch.

Kiddle: the most-recommended and the least accountable

Kiddle is the best-known kids search engine in the English-speaking world, routinely listed on school pages, library guides and parenting sites. It is a good worked example precisely because it is typical in its architecture and unusual in its opacity.

What it is. A Google Programmable Search Engine instance with Google SafeSearch enabled, plus Kiddle's own block-lists and hand-picked results, presented in large type with big thumbnails and a robot mascot. Its own About page states that Kiddle "is powered by Google Safe Search but is not affiliated with Google LLC" (kiddle.co/about.php, checked 19 August 2026). Its results are arranged in tiers: positions one to three are described as safe pages written for children and hand-picked by Kiddle editors; four to seven as trusted sites in simple language; eight and below as ordinary adult-facing sites filtered by Google SafeSearch. It also runs a children's encyclopedia, Kpedia, built largely on openly licensed Wikipedia-derived material.

Who runs it is not disclosed. The site names no company, no legal entity, no editorial team and no leadership, and its domain registration sits behind a privacy proxy. Reporting since 2016 has pointed to a Russian entrepreneur named Vladislav Golunov, first surfaced by EdSurge on 1 March 2016 and corroborated by The Register on 11 April 2016 through source-code tracking tags and domain records. That attribution has never been confirmed by Kiddle, and a decade later no accountable operator has been established. For a product handed to children in schools, that is the fact that governs everything else: there is nobody to name in a data-processing agreement, nobody to send a safeguarding complaint to, and no jurisdiction in which to make one.

It is not a Google product. Several outlets reported it as one when it appeared in early 2016 and later published corrections. The belief has never gone away, and it is the reason many adults trust the site.

The documented record, with dates

These are the events on the public record. They are given with sources and dates because the category is full of recycled listicle copy that repeats none of them.

1 March 2016 — the block-list controversy. PinkNews and others reported that Kiddle blocked queries including gay, lesbian, bisexual, transgender and intersex, returning a message saying it was "hard to guarantee the safety of all the search results for such queries" and suggesting the child talk to a parent or guardian. The Next Web reported the same day that the list also covered sex, menstruation, circumcision, death, birth, self harm and suicide.
1 March 2016 — the bypass. The Next Web demonstrated that slight misspellings or added spaces defeated the block-list entirely, and criticised Kiddle for returning no help or support on self-harm queries where Google returned crisis resources.
Around 2 March 2016 — after the backlash, terms including gay and puberty were unblocked, per CBC News. Sources disagree about how much was restored; Gay Times published a piece arguing the "blocks all LGBT content" framing overstated it. The underlying design — an undisclosed block-list maintained by anonymous editors, changeable at any time without notice — was not changed.
11 April 2016 — The Register reported a page at kiddle.co/redir.php that handed users through to Lukol, an anonymity-branded search engine: an unfiltered search path sitting on a child-safety domain. Kiddle's support said it had been used for internal testing during beta and was not linked from any Kiddle page.
July 2023 — Spanish-language search added. September 2025 — logo redesigned.
21 May 2026 — City Journal published an investigation by Ashley Rindsberg reporting that the operator remains unknown, and describing what it presents as systematic slant in Kiddle's child-facing encyclopedia content: Russia's invasion of Ukraine framed as a "military operation," a Stalin entry emphasising the building of a "strong, modern nation," terrorism mentioned once and at the foot of the Hamas article, no mention of the camps in the Uyghur article, and no discussion of Covid-19 origins in the Wuhan Institute of Virology entry. It also reported ads being served to child users and privacy-policy language about using cookies "to personalize content and ads," which the article argued conflicts with Google's own rules for child-directed content. Neither Kiddle nor Google responded to its requests for comment. City Journal is published by a think tank with an editorial viewpoint and these findings are one outlet's; they are also specific, dated and checkable by anyone who opens the pages.

Kiddle's own position is that it collects no personally identifiable information and deletes its logs every 24 hours. There is no way for a parent to verify that, because there is no identified company, no jurisdiction and no audit.

How to judge one of these products

The useful questions are not about features. They are about accountability and mechanism, and there are four of them.

Who operates it, and under which country's law? If the site does not name a company and a jurisdiction, every other claim it makes is unverifiable in principle. This is the question that eliminates most of the category.
Whose index is underneath, and is that disclosed? Almost always Google or Bing. A product that will not say is telling you something.
What happens on the hard queries? Type a self-harm or sexual-health query yourself, before a child does. Block message, crisis resources, or ordinary results — the answer tells you what the filter is actually for.
Who maintains the block-list, and can you see it? Unnamed editors making undisclosed editorial decisions about what children may read is a governance question, not a technical one.

Structurally there are three other routes, and they trade convenience for accountability in different ways. SafeSearch enabled directly on the underlying engine, on a supervised child account managed by an adult, gives you the same filter these sites resell, without an unidentified intermediary — but no kid-friendly presentation and no curated top results. Network- or device-level filtering from a named vendor covers everything the child does rather than one search box, and comes with a company, a contract and a published privacy policy — but it is a purchase and an administration job. An engine with a permanent adult-content filter is a third pattern: Swisscows, a Swiss service operated by Hulbee AG, serves Bing's results with the pornography filter always on and not switchable, which is a real product decision rather than a toggle. It is not designed for children specifically, and the permanent filter also degrades legitimate medical and sexual-health queries, which is exactly the trade-off you are accepting.

The last point is the one that survives every product comparison: filtering is not supervision. Every mechanism described on this page is defeatable, and the documented failures in this category have not mostly been failures to block pornography. They have been unaccountable operators, block-lists that shut down the questions children most need answered, and slant in material that passed every filter.

Frequently asked questions

Is Kiddle Google's search engine for kids?

No. Google does not own, operate or endorse Kiddle, and Kiddle's own About page states it is "not affiliated with Google LLC." It is an independent site built on Google's Programmable Search Engine product with SafeSearch enabled. The confusion dates from early 2016, when several outlets reported it as a Google product and later published corrections.

Do kids search engines have their own index?

Almost never. The overwhelming majority are filtered front ends over Google or Microsoft Bing, using SafeSearch plus their own domain and keyword block-lists. Building a web-scale index is prohibitively expensive, so what a kids search engine actually supplies is the filter, the presentation and a small number of hand-picked results, not the search.

Are kids search engines safe?

They are filtered, which is not the same thing. Keyword block-lists can be defeated by misspelling a word or adding a space, only a handful of top results are typically checked by a human, and a domain that is approved overall can host a page that is not. Filtering also cannot detect inaccuracy or bias in results that pass it.

Who owns Kiddle?

Nobody knows publicly. The site names no company, no editors and no jurisdiction, and its domain registration is masked by a privacy proxy. Reporting since 2016 has associated it with a Russian entrepreneur, Vladislav Golunov, but that has never been confirmed and no legal operating entity has been identified in the decade since.

What did Kiddle block in 2016?

On 1 March 2016 it was reported to block queries including gay, lesbian, bisexual, transgender and intersex, returning a message suggesting the child speak to a parent. The Next Web reported the list also covered sex, menstruation, circumcision, death, self harm and suicide. Some terms, including gay and puberty, were unblocked within about a day after the backlash.

What should a school ask before deploying a safe search engine?

Four things: who operates it and under whose law, whose index supplies the results, what the product returns for a self-harm or sexual-health query, and who maintains the block-list. A service that cannot answer the first question cannot be held to a data-processing agreement or a safeguarding process, whatever its filtering claims.

What is the difference between SafeSearch and a kids search engine?

SafeSearch is a setting on Google or Bing that suppresses explicit content using that engine's own classifiers. A kids search engine is usually a third-party site that switches SafeSearch on for you, adds its own block-lists and a child-friendly design, and places itself between the child and the engine. The filtering underneath is largely the same.

Kids and safe search engines: what they really are