Some New Data on Age Authentication Balk Rates

October 14, 2026 · by · in Content Regulation

Created by ChatGPT May 2026

The issue is: if a website imposes a government-compelled age authentication wall, what percentage of users will stop at the authentication wall, either because the users turn around or fail to navigate the process? I call that percentage the “balk” rate. For background on prior studies and why balk rates can vary widely, read my prior coverage first.

I’m now reporting on a study by All About Cookies entitled “72% of Americans Don’t Trust the Companies Verifying Their Age Online.” The writeup qualifies the headline claim: “72% of Americans express low levels of trust in the companies and services handling age verification data, including 40% who do not trust them at all.” The other 32% say they “slightly” trust the authenticators.

This chart shows what happened when users were confronted with an age authentication wall (users could select multiple options):

I disregard the first and third options because they are self-reporting, and that isn’t an “authentication” at all. Instead, focus on the bottom three options, all of which are rejections of the age authentication wall (go somewhere else, abandon the effort, or bypass the wall using VPNs). The first two are balk rates; the third also rejects the age authentication wall.

One way of reading this data is that only 31% of consumers were willing to age-authenticate; all other consumers chose some other non-authentication option if they could. Because users could pick multiple choices, we can’t combine the three rejection options (which adds up to 32%) and compare it to the 31% who said they authenticate. But even if we take the most conservative reading of this data and say that there’s at least an 11% balk rate, that’s still a lot of disruption of existing user activity.

The report summarizes this chart with this confusing statement: “Age verification laws don’t get more people to verify: 31% complete verification in states with and without laws. Instead, people in law states are 7x more likely to give up entirely and 4x more likely to use a VPN.” The survey didn’t provide the supporting data to explain the 7x assessment, but I interpret it as a warning that the quantity of consumer balks will soar as more states’ age authentication mandates come into effect.

The next chart asks how comfortable users would be sharing the specific items of data to get through an age authentication wall:

As you can see, the “least” troubling option is an automatically scanned selfie, which more than 2/3 said they would be uncomfortable with sharing. Credit card authentication–which used to be the benchmark for age authentication in the 1990s and is now considered no longer credible–made nearly 9 of 10 consumers uncomfortable.

Over time, these discomfort rates might organically decline as age authentication mandates become ubiquitous, which will cause consumers become to desensitized to the omnipresent privacy invasions.

Zero-knowledge proofs might also change how consumers feel about sharing their government-issued IDs because the ID contents would not be shared. However, to reach that point, consumers first will have to overcome the many troubling implications of installing a government-issued digital ID on their device. Personally, I’m a hard “no” on digital IDs.

We can’t directly infer a balk rate from this data, but it’s not good. When users are uncomfortable with the data disclosure, they weigh the costs/benefits of proceeding. If they can’t easily assess the benefits–which will often be the case with new services that the consumer doesn’t know well–then the discomfort is likely to increase balk rates. As I mention in my Segregate-and-Suppress paper, the resulting changes in consumer behavior are likely to hinder new market entrants and overall market dynamism.

Overall, I don’t see how anyone, especially government regulators, can look at the emerging balk rate data and think that the Internet is going to be just fine when online age authentication mandates become ubiquitous.

Blog Posts on Segregate-and-Suppress Obligations