Yesterday I was asked to verify my identity using an AI-powered video verification process.
If you haven’t encountered one of these yet, here’s how they work. You take a video of yourself while following prompts on the screen and, if everything goes well, the system decides you’re a real person.
Sometimes the prompts are simple. One process I encountered just wanted my face fully in view and gave me instructions to move my phone closer or farther away. Others are more complicated. They might ask you to match the position of a face shown on the screen, align your nose with a moving dot, or keep your eyes open and looking directly at the camera.
These systems are becoming increasingly common. Online dating apps use them because, unfortunately, people suck, and platforms need ways to reduce scams, fake profiles, and other bad behavior. If you’ve followed my recent adventures in online dating—and yes, those apps can be used to find friends as well as romance—you know I’ve encountered quite a few of these systems.
But this isn’t just a dating app problem.
Want to open a bank account online? You may have to complete a video verification process.
Applying for a credit card? Same thing.
The federal government uses similar technology when you create accounts with agencies like the IRS and the Social Security Administration. Whether we like it or not, all of us are likely to encounter these systems eventually.
I first ran into this issue years ago as a software engineer working for a financial services company.
As a customer, employee, software engineer, and someone with a background in security, I was able to successfully advocate for accessibility when my employer adopted video identity verification. It wasn’t easy. Many people believed accessibility and security were competing goals.
Part of that concern came from guidance published by the National Institute of Standards and Technology (NIST). Government agencies, banks, and other security-sensitive organizations often look to NIST standards when designing identity verification systems. Because I understood those standards, I was able to make the case that accessibility and security could coexist.
I was also fortunate. As an employee, I had direct access to decision-makers.
Most people don’t.
Oh, and did I mention that I’m blind?
Because I’m blind, many of these verification systems are nearly impossible for me to use independently.
I can’t see where my eyes are looking. I often have difficulty keeping them open. Instructions like, “Keep your eyes open and look directly at the camera,” aren’t merely difficult—they’re impossible.
Likewise, I can’t follow instructions such as, “Align the tip of your nose with the dot on the screen,” because I can’t see the dot.
There are workarounds.
I can ask someone to help me. I’ve done that when I wanted or needed something badly enough.
I can use visual interpretation services such as Aira or Be My Eyes. Aira, in particular, allows an agent to view both my camera feed and my screen, making it possible for them to guide me through some processes.
Even then, success isn’t guaranteed.
Aira and I once spent nearly an hour trying to complete an ID verification process before finally giving up.
On the other hand, I’ve encountered systems that simply say things like, “Move your phone closer,” or, “Move your phone farther away.” Those systems have often worked well because they provide feedback I can actually use.
I’ve also had companies accept photos of the front and back of my state ID instead. Those processes are usually accessible, but they’re becoming less common because photos are easier to fake.
Some organizations take a more flexible approach.
Government agencies, banks, and other large organizations often use multiple methods to establish confidence in your identity. They might ask for your address, the last four digits of your Social Security number, or information from your credit history. Based on your responses, they calculate how likely it is that you’re really you.
If the system is confident enough, you may never need video verification at all.
And if you do, many organizations offer a fallback option where you can connect with a real person over video chat.
There are also nontechnical solutions.
You can visit your local Social Security office. If you’ve ever done that, you know it’s a thrilling way to spend a Wednesday morning.
You can visit your bank, which is slightly more enjoyable because bank employees are generally friendlier.
But I’m a technology person, so I’m interested in how we can make technology work better.
When a process like this fails, the first thing I want is the ability to contact a real person. I want help completing the task, and I want the company to know that a problem exists.
After all, if nobody tells them, how will they ever improve?
Yesterday I encountered yet another inaccessible verification system.
Before contacting support, I checked the app’s accessibility statement. It claimed compatibility with VoiceOver, the screen reader built into the iPhone that makes it possible for me to use the device independently.
Armed with that information, I filled out their support form, explained the issue, and attached photos of my ID.
I’ve solved similar problems that way before.
A few minutes later, I received an automated response informing me that someone would get back to me within 48 hours.
Think about that for a moment.
A sighted customer could complete the process in two minutes.
I might have to wait two days.
Do you find that acceptable?
I don’t.
Fortunately, someone replied about three hours later.
Unfortunately, the response was clearly a form letter explaining how to use the verification system I’d already explained I couldn’t use.
I replied, perhaps a little bluntly, explaining that I didn’t think anyone had actually read my original message. I reminded them that they claimed accessibility and that I had already spent money to access their service.
This morning I woke up to a much better response.
They offered a manual verification process.
It still required a video, but this time they provided instructions I could actually follow.
They asked me to:
- State a verification phrase they provided.
- Hold up an object and describe it using one word.
- Continue speaking while turning my head left and right.
It’s not ideal.
But it’s something I can do.
And it’s a lot better than a dead end.
Still, the entire experience left me with a familiar feeling.
The feeling that I’m forcing myself to belong.
The feeling that I’m being tolerated rather than welcomed.
That’s the difference between accessibility and inclusion.
Accessibility gets me through the door.
Inclusion makes me feel like I was invited.
Here’s what I imagine an inclusive verification process might look like.
VoiceOver could say, “Your light isn’t on. Please turn on a light so I can see your face.”
Then, “Place your phone in front of your face and slowly move it away.”
As I move the phone, it could provide real-time feedback.
“Farther… farther… farther… stop.”
Then:
“Move your phone down.”
“Good.”
“A little farther.”
“Stop.”
Then:
“Move your phone left.”
“Good.”
“Stop.”
Finally:
“Your face is centered. Slowly turn your head left.”
“Good.”
“Stop.”
“Now turn your head right.”
“Good.”
“Stop.”
“Congratulations. Verification complete.”
With the proper directions and feedback, I could complete the same process as a sighted person, independently.
Before anyone says that’s too difficult or too expensive, much of this technology already exists. Similar capabilities are built into modern smartphones today. They simply need to be combined in thoughtful ways.
And remember, most companies aren’t building these systems themselves. They’re purchasing them from specialized vendors. Improvements made by those vendors benefit thousands of organizations and millions of customers.
Of course, inclusion doesn’t stop with blind people.
Someone with limited mobility might need a different process. Someone with a speech disability might need another.
No single solution will work for everyone.
That’s why every system needs an immediate path to a human being.
Not in 48 hours.
Not after five emails.
Immediately.
A secure video call with a trained support representative could solve many of these problems in minutes.
Technology should be the first option.
Human support should be the safety net.
Both matter.
Artificial intelligence has come a long way, but it’s still surprisingly stupid.
At the same time, we’ve already seen that AI can become remarkably good at specific tasks when trained properly.
I’ve started receiving sales calls from AI systems. I can usually tell because their speech patterns are just a little off and their responses take just a little too long. Sometimes, when I’ve asked directly, they’ve even admitted they’re AI.
But they’re getting better.
Fast.
Today I can ask ChatGPT to describe a photograph in a way that’s useful to me as a blind person.
Tomorrow, AI systems will become better at understanding context, disability, and individual needs.
They’ll learn that some people can’t keep their eyes open.
Some people don’t have eyes at all.
Some people can’t turn their heads.
Some people communicate differently.
AI has the potential to make these systems dramatically more inclusive while also making identity verification faster, safer, and less expensive.
But it won’t happen automatically.
It will happen because people encounter barriers, speak up, and refuse to accept being excluded.
That’s why we need to keep talking about these problems.
That’s why designers need to hear these stories.
And that’s why accessibility isn’t enough.
Inclusion means making every customer feel like they’re wanted, not just tolerated.
It’s reasonable.
It’s possible.
And it’s the right thing to do.
What do you think?
Have you ever encountered a piece of technology that worked well for you but created barriers for someone else? Do you have ideas about how technology could be made more inclusive?
Maybe those ideas are possible today. Maybe they’re only a few years away. Either way, I’d love to hear them. Share your thoughts in the comments.
And if you’re reading this through my email newsletter, you don’t even have to visit the website. Just reply to the email. Your reply becomes a comment. That’s one small way I’ve tried to make this space more inclusive and accessible for everyone.
Finally, if you appreciate my writing, want me to keep doing more of it, and agree that writers deserve to be paid for their work, consider becoming a Supporting Member. To make that as inclusive as possible, you choose what you pay. Give as much or as little as you’re able and feel comfortable contributing.
Whether you read, comment, share, subscribe, or support financially, thank you for being here. You’re helping create the kind of community where people are welcomed, not merely tolerated.
Leave a Reply