Age-gating laws block adults and youth alike from freely accessing services on the web and force users to trade their anonymity—a pillar of online expression—for a system in which they are bound to their real-life identities. This surveillance regime stretches beyond just age restrictions on certain content; much of this infrastructure is also connected to government plans for creating a digital system of proof of identity.

So how does age gating actually work? The age and identity verification industry has devised countless different methods platforms can purchase to—in theory—figure out the ages and/or identities of their users.  But in practice, there is no technology available that is entirely privacy-protective, fully accurate, and that guarantees complete coverage of the population. Full stop.

Every system of age verification or age estimation demands that users hand over sensitive and oftentimes immutable personal information that links their offline identity to their online activity, risking their safety and security in the process.

With that said, as we see more of these laws roll out across the U.S. and the rest of the world, it’s important to understand the differences between these technologies so you can better identify the specific risks of each method, and make smart decisions about how you share your own data.

Age Assurance Methods

There are many different technologies that are being developed, attempted, and deployed to establish user age. In many cases, a single platform will have implemented a mixture of methods. For example, a user may need to submit both a physical government ID and a face scan as part of a liveliness check to establish that they are the person pictured on the physical ID. 

 Age assurance methods generally fall into three categories:

  1. Age Attestation
  2. Age Estimation
  3. ID-bound Proof

Age Attestation

Self-attestation 

Sometimes, you’ll be asked to declare your age, without requiring any form of verification. One way this might happen is through one-off self-attestation. This type of age attestation has been around for a while; you may have seen it when an alcohol website asks if you’re over 21, or when Steam asks you to input your age to view game content that may not be appropriate for all ages. It’s usually implemented as a pop-up on a website, and they might ask you for your age every time you enter, or remember it between site accesses. This sort of attestation provides an indication that the site may not be appropriate for all viewers, but gives users the autonomy and respect to make that decision for themselves.

An alternative proposed approach to declaring your own age, called device-bound age attestation, is to have you set your age on your operating system or on App Stores before you can make purchases or browse the web. This age or age range might then be shared with websites or apps. On an Apple device, that age can be modified after creation, as long as an adult age is chosen. It’s important to separate device-bound age attestation from methods that require age verification or estimation at the device or app store level (common to digital ID solutions and some proposed laws). It’s only attestation if you’re permitted to set your age to whatever you choose without needing to prove anything to your provider or another party—providing flexibility for age declaration outside of mandatory age verification.

Attestation through parental controls

The sort of parental controls found on Apple and Android devices, Windows computers, and video game consoles provide the most flexible way for parents to manage what content their minor children can access. These settings can be applied through the device operating system, third-party applications, or by establishing a child account. Decisions about what content a young person can access are made via consent-driven mechanisms. As the manager, the parent or guardian will see requests and activity from their child depending on how strict or lax the settings are set. This could include requests to install an app, make a purchase on an app store, communicate with a new contact, or browse a particular website. The parent or guardian can then choose whether or not to accept the request and allow the activity. 

One survey that collected answers from 1,000 parents found that parental controls are underutilized. Adoption of parental controls varied widely, from 51% on tablets to 35% on video game consoles. To help encourage more parents to make use of these settings, companies should continue to make them clearer and easier to use and manage. Parental controls are better suited to accommodating diverse cultural contexts and individual family concerns than a one-size-fits-all government mandate. It’s also safer to use native settings–or settings provided by the operating system itself–than it is to rely on third-party parental control applications. These applications have experienced data breaches and often effectively function as spyware.

Age Estimation

Instead of asking you directly, the system guesses your age based on data it collects about you.

Age estimation through photo and facial estimation

Age estimation by photo or live facial age analysis is when a system uses an image of a face to guess a person’s age.

A poorly designed system might improperly store these facial images or retain them for significant periods, creating a risk of data leakage. Our faces are unique, immutable, and constantly on display. In the hands of an adversary, and cross-referenced to other readily available information about us, this information can expose intimate details about us or lead to biometric tracking.

This technology has also proven fickle and often inaccurate, causing false negatives and positives, exacerbation of racial biases, and unprotected usage of biometric data to complete the analysis. And because it’s usually conducted with AI models, there often isn’t a way for a user to challenge a decision directly without falling back on more intrusive methods like submitting a government ID. 

Age inference based on user data and third party services

Age inference systems are normally conducted through estimating how old someone is based on their account information or querying other databases, where the account may have done age verification already, to cross reference with the existing information they have on that account.

Age inference includes but not limited to:

In order to view how old someone is via account information associated with their email, services often use data brokers to provide this information. This incentivizes even more collection of our data for the sake of age estimation and rewards data brokers for collecting a mass of data on people. Also, regulation of these age inference services varies based on a country’s privacy laws.

ID-bound Proof

ID-bound proofs, methods that use your government issued ID, are often used as a fallback for failed age estimation. Consequently, any government-issued ID backed verification disproportionately excludes certain demographics from accessing online services. A significant portion of the U.S. population does not have access to government-issued IDs, with millions of adults lacking a valid driver’s license or state-issued ID. This disproportionately affects Black Americans, Hispanic Americans, immigrants, and individuals with disabilities, who are less likely to possess the necessary identification. In addition, non-U.S. citizens, including undocumented immigrants, face barriers to acquiring government-issued IDs. The exclusionary nature of document-based verification systems is a major concern, as it could prevent entire communities from accessing essential services or engaging in online spaces.

Physical ID uploaded and stored as an image 

When an image of a physical ID is required, users are forced to upload—not just momentarily display—sensitive personal information, such as government-issued ID or biometric identifiers, to third-party services in order to gain access to age-restricted content. This creates significant privacy and security concerns, as users have no direct control over who receives and stores their personal data, where it is sent, and how it may be accessed, used, or leaked outside the immediate verification process.

Requiring users to digitally hand over government-issued identification to verify their age introduces substantial privacy risks. Once sensitive information like a government-issued ID is uploaded to a website or third-party service, there is no guarantee that it will be handled securely. The verification process typically involves transmitting this data across multiple intermediaries, which means the risk of a data breach is heightened. The misuse of sensitive personal data, such as government IDs, has been demonstrated in numerous high-profile cases, including the breach of the age verification company AU10TIX, which exposed login credentials for over a year, and the hack of the messaging application Discord. Justifiable privacy and security concerns may chill users from accessing platforms they are lawfully entitled to access.

Device-bound digital ID

Device-bound digital ID is a credential that is locally stored on your device. This comes in the form of government or privately-run wallet applications, like those offered by Apple and Google. Digital IDs are subject to a higher level of security within the Google and Apple wallets (as they should be). This means they are not synced to your account or across services. If you lose the device, you will need to reissue a new credential to the new one. Websites and services can directly query your digital ID to reveal only certain information from your ID, like age range, instead of sharing all of your information. This is called “selective disclosure."

There are many reasons someone may not be able to acquire a digital ID, preventing them from relying on this option. This includes lack of access to a smartphone, sharing devices with another person, or inability to get a physical ID. No universal standards exist governing how ID expiration, name changes, or address updates affect the validity of digital identity credentials. How to handle status changes is left up to the credential issuer.

Asynchronous and Offline Tokens

This is an issued token of some kind that doesn’t necessarily need network access to an external party or service every time you use it to establish your age with a verifier when they ask. A common danger in age verification services is the proliferation of multiple third-parties and custom solutions, which vary widely in their implementation and security. One proposal to avoid this is to centralize age checks with a trusted service that provides tokens that can be used to pass age checks in other places. Although this method requires a user to still submit to age verification or estimation once, after passing the initial facial age estimation or ID check, a user is issued a digital token they can present later to to show that they've previously passed an age check. The most popular proposal, AgeKeys, is similar to passkeys in that the tokens will be saved to a device or third-party password store, and can then be easily accessed after unlocking with your preferred on-device biometric verification or pin code.

Lessons Learned

With lessons pulled from the problems with the age verification rollout in the UK and various U.S. states, age verification widens risk for everyone by presenting scope creep and blocking web information access. Privacy-preserving methods to determine age exist such as presenting an age threshold instead of your exact birth date, but have not been mass deployed or stress tested yet. Which is why policy safeguards around the deployed technology matter just as much, if not more. 

Much of the infrastructure around age verification is entangled with other mandates, like deployment of digital ID. Which is why so many digital offerings get coupled with age verification as a “benefit” to the holder. In reality it’s more of a plus for the governments that want to deploy mandatory age verification and the vendors that present their implementation that often contains multiple methods. Instead of working on a singular path to age-gate the entire web, there should be a diversity of privacy-preserving ways to attest age without locking everyone into a singular platform or method. Ultimately, offering multiple options rather than focusing on a single method that would further restrict those who can’t use that particular path.