Google will guess Canadians' age from their searches and YouTube history; adults flagged as minors must show ID or a selfie

15.09.2026 8 min 23

Google will start guessing the age of its Canadian users from what they search for and which YouTube videos they watch. In a blog post on Monday, 14 September, Jeanette Patell, the company's head of government affairs in Canada, said the machine-learning "age assurance" system that launched in the United States last year will "in the coming days" begin a gradual rollout in Canada. Accounts the model judges to belong to someone under 18 get a changed set of defaults: no personalised ads, no location Timeline in Maps, no adult-rated apps on Google Play, bedtime reminders on YouTube. An adult caught by mistake can get the settings back by uploading a government ID or a selfie. Google says it collects nothing new; it draws conclusions from what it already holds.

In brief

  • The model reads "signals already associated with a user's account, such as the types of information a user has searched for or the categories of videos they've watched on YouTube" and decides whether the person is likely over or under 18. There is no switch to refuse the estimate.
  • Flagged users are told by email and inside Google products; the protections are the same as in the US rollout of July 2025. Getting them lifted means proving age with a photo of a government document or a selfie.
  • The timing follows Bill C-34, the Safe Social Media Act tabled on 10 June, which would set 16 as the minimum age for social media accounts and leave the method of checking to be worked out with platforms. Google's line is that responsibility sits with each service, not with the app store, as Australia has just decided.
  • The estimate is made on the account, not the connection, so a VPN changes nothing; and the appeal route sends an ID or a face to a company whose industry has leaked 2.15 billion identity records in 88 breaches.

What Google says it will do

The post describes two stages. First, estimation: "Our age estimation model uses machine learning to interpret a variety of signals already associated with a user's account," and "does not collect any new information not already associated with the user's account." Second, verification, only when the estimate is challenged: "If we incorrectly estimate a user to be under 18, the user has the option to correct their age, including by uploading a photo of their government ID or a selfie." For accounts the model places under 18, Google lists four changes: YouTube's digital wellbeing tools switch on, with reminders to take a break and go to bed and "safeguards to content recommendations, including limiting repetitive views of some kinds of content"; Timeline in Maps is disabled; personalised advertising is off and age-sensitive ad categories are restricted; and apps rated for adults on Google Play become unavailable. Patell says the rollout will be gradual, "carefully monitoring progress as we expand its availability". None of this is new technology: TechCrunch reported the same system, with the same list of consequences and the same ID-or-selfie appeal, when Google began testing it in the United States on 31 July 2025.

18the line the model draws: over or under, from search and viewing history
0new data collected, according to Google; the estimate runs on what the account already contains
16minimum age for social media accounts in Bill C-34, introduced on 10 June 2026 and not yet law
3% or $10 Mthe higher of the two, the top penalty the bill proposes for platforms that fail their duties

Why now: Bill C-34 and the fight over who checks

The Safe Social Media Act, introduced by the Liberal government on 10 June, would create a Digital Safety Commission and require social media services and AI chatbots to assess risks, publish safety plans and act on seven categories of harmful content within 24 hours of a flag. Its headline measure is a minimum age of 16 for social media accounts, with the commission able to exempt a platform that shows "sufficient safeguards for children". How age would be checked is not in the bill; the government has said it will consult platforms on methods. That is the gap Google is filling on its own terms. Patell's post argues that "liability and responsibility rest with every service owner", a direct answer to the alternative Australia adopted this month, where the age check happens once at the app store and is passed down to apps, and to the British model, which lists seven acceptable methods from facial estimation to bank checks. Meta got there first in Canada: it already uses AI, including "visual analysis" of photos and videos since May, to move suspected minors into teen accounts, while confirming that the child-safety measures it accepted in its US legal settlement will not apply north of the border.

What the model can and cannot see: the signals are on the account, so they follow you across devices and past any VPN or private browsing that is still signed in. They do not exist for a signed-out session, and they are thin for an account used mainly for mail. A history of gaming videos, school-related searches or anime is exactly the kind of pattern that will put some adults on the wrong side of 18, and Google's own text presents the ID upload as the fix.

What a flagged adult gives up, and what to do

The changes are quieter than a ban and worth listing. Losing personalised ads costs nothing. Losing Timeline means Maps stops keeping your location history, which many people would consider a gain. Losing adult-rated apps on Play and gaining bedtime nags on YouTube is an irritant. The real cost is the appeal: to be treated as an adult again you hand Google a passport, driving licence or a facial image, on the word of a model whose error rate the company does not publish. The same logic is arriving at the operating-system level in Windows, where apps will be told a user's age bracket, and the sum of these systems is a web where age is inferred everywhere and proven, with documents, whenever the inference is wrong. For Canadians who want to keep their options, the practical steps are small: watch for the email Google says it will send; decide whether the four restrictions matter to you before uploading anything; keep the account you use for search and YouTube separate from the one that holds your documents and payments; and remember that a signed-out browser gives the model nothing to read. The larger decision, whether platforms guess, app stores check or the state issues a credential, is the one Bill C-34 leaves open, and Google has just cast its vote.

Does this apply to everyone with a Google account in Canada?
Google says the rollout starts "in the coming days" and expands gradually. The model runs on signed-in accounts and uses signals already tied to them; it has no data to work with for signed-out use.
Will Google collect new data to guess my age?
Google says no: the model interprets what is already associated with the account, such as search topics and YouTube video categories. What changes is what Google does with that data.
What happens if the model decides I am under 18?
You get an email and in-product notices. Personalised ads and age-sensitive ad categories are switched off, Timeline in Maps is disabled, adult-rated apps on Google Play are blocked, and YouTube turns on break and bedtime reminders and limits repeated recommendations of some content.
How do I prove I am an adult?
By uploading a photo of a government ID or a selfie through the notice Google sends. There is no other route described in the post, and no way to opt out of the estimate itself.
Is this required by Canadian law?
Not yet. Bill C-34 would set a minimum age of 16 for social media accounts and create a regulator, but it has not passed and does not specify how age must be checked. Google is moving ahead of it and arguing that each service should do its own checking rather than the app store or a government system.

googlecanadayoutubeage estimationage verificationprofilingminorssocial mediamachine learningdigital idprivacymetabill c-34

Read also