I was sending a message to my friend in Vietnam who is a professor of social work, something innocuous about meeting up in the near future and maybe visiting the city of Hue together (the coffee would be on me). Within a second of sending her the Facebook message my screen froze and I was informed by FB that I needed to prove I was a human being. But, it wasn’t one of those easy, straight-forward CAPTCHA things.
I needed to create and forward a video selfie. Amazed that FB wanted me to perform in order to keep my account (I was wondering – song and dance? poetry recital? reading the Constitution backwards in a MAGA hat?); I wondered what they would take as proof of my humanity.
My video camera flickered on and they started giving me directions on how to do my FB-saving performance. They said “Move your head up!” so I moved my head up, then looked into the camera, flipped them the bird (with both hands) and mouthed something that would severely hurt this article in Google rankings if I typed it here.
“Move your head to the left!” Same thing. After 4 or 5 commands I was informed, immediately, that I was a robot and my FB account was now permanently deleted. If flipping the bird couldn’t establish my humanity, what could?
To try to understand why FB challenged me in the first place, I theorized that it might be due to the VPN I sometimes use. In fact, I often use a VPN when I am online, because I am currently living and traveling through Asia. If I am where several people use the same Wi-Fi, not to use a VPN would risk having my information stolen. Sometimes Gmail will even ask me to re-verify my identity because of a suddenly new IP address, but this is always a very simple, fair and straight-forward procedure.
By not using tried and true “CAPTCHA” and by obviously using a flawed AI system, and not allowing any appeals process to a human, it seemed that Facebook might have even wanted my harmless and non-robotic account of 15 years to disappear. But why? Well, I soon discovered a reason.
Now first, I did what any angry American with an axe to grind does, I went to Reddit. Sure enough, I found a couple subreddits dealing with this problem – I was not the only person with this experience, by far. I discovered that the BBC actually did a story, back in July, about how Facebook seemed to be deleting accounts of real, innocent people while expressing total denial. That was four months ago, and they are still doing it.
It turns out that lots of people are losing 15 to 20 or more years of family photos, and small business people are losing money, and, apparently, their businesses. Lots of people are getting hurt.
There was a petition started by a lady in Ontario now with 50,000 signatures (so far) of verified human beings (real carbon-based organisms) who had their FB accounts deleted for, apparently, no discernible reason.
So why do I bring this to your attention? Because it seems clear that Facebook is proving what corporate America intends to do now that it has its grubby, greedy hands on flawed and shoddy AI systems. The systems don’t work very well, but they work well enough to save money (while hurting people). Facebook has seemingly tipped the hand of corporate America: they are going to use AI to save/make money, even if it hurts people.
But, as someone on Reddit asked, “How foolish can they be? They are wiping out tens of thousands of potential money-makers.” Here’s the answer.
The AI “human verification” system is broken, and it’s apparent they know it. It misreads faces, or lighting, or skin tones, or eye movements, or, apparently, people who flip the bird (the most basic way to demonstrate one’s humanity). It fails a lot, and often nukes perfectly benign accounts. FB will never admit this, because admitting it means liability.
It seems they’re deleting accounts because it’s cheaper than supporting them. You see, Facebook has a problem, it got loaded with bots and spammers. So, hypothetically, what should a company do in this situation? Hire people (it’s really good for the economy) and let them review the shoddy, piece-of-junk job the AI system is doing. If a human being sees another human being flipping the bird, that account is good.
But, Meta, you don’t really want folks to prove their humanity, do you? You want to get rid of the bots and spammers that infiltrated your system, as cheaply as possible, and this is also taking tens of thousands of small business owners and innocent people along as collateral damage. It would cost too much to save these innocent people, wouldn’t it?
Human moderation costs money. AI purges are free. When in doubt, the algorithm deletes a person. Facebook gets to brag about fighting “bots” while wiping out thousands of real users who had put their trust in the company. They wanted people to rely on their product, people did and then Meta pulled the carpet out from under them for their loyalty and support.
Facebook also hides behind the AI. First of all, they don’t let you complain. When they wipe out your account, they also tell you that there are no appeals (based on my experience). If you were allowed to complain, they might say: “We reviewed your case and found no error.” Except, no human reviews anything.
AI makes the mistake; AI rubber-stamps the mistake; the company then says “Sorry, AI made the mistake! We’re only human, forgive us!” Actually, they don’t even have the decency to say that. They say nothing. Just keep denying!
AI becomes the fall guy for corporate negligence, while corporate America actually wanted it that way. There is no government oversight to stop this type of abuse so memories and businesses become obliterated.
They’re possibly monetizing vulnerability. It is completely possible that Facebook trains its AI (inadvertently) on biased data, then punishes the same groups already marginalized in society. Facebook’s own Civil Rights Audit (2020) found that its machine learning systems had fairness and discrimination risks. Based on what I’m seeing on Reddit, it looks as if lots of older folks got their accounts obliterated. They paid their dues to society, contributed, earned retirement and Facebook wipes out their stored memories.
This will not be isolated. Facebook is just the first major corporation to lean fully into AI cruelty. We are now seeing what all the experts warned us about – AI being used to harm people. This is an acceptable business model these days. Facebook seems to have the dubious honor of being the first to do this on a grand scale, and they are seemingly brazen about it.
Unless folks in Congress do something, we’ll see the first examples of negative AI fallout through this misadventure. Facebook will set a precedent which will haunt innocent people well into the future. AI is not ready to do good things for masses of people, but it is ready to do harmful things. If companies can make money from the harmful things, they will.
AI shadowbans people based on misreading jokes, sarcasm or political speech. Humans understand context; AI doesn’t. You quote a news article about violence? AI flags you for violence.
AI-driven facial recognition mislabels millions, especially non-white users. And guess what? Errors disproportionately target Black, Asian, South Asian and Indigenous users.
AI can auto-flags disabled users as “bots” because of atypical movement patterns. People with Parkinson’s, tremors, stroke aftereffects, can be banned automatically.
AI deletes small businesses that “violate commerce rules” when they didn’t. Thousands of them. Lost income, zero recourse.
AI decides whether your political post stays up or vanishes. This is power no corporation should have, and they’re giving it to flawed, malfunctioning, garbage software.
Some people on Reddit asked: “How foolish can Facebook be? They’re deleting tens of thousands of potential money-makers.” I want to elaborate on the answer I partially gave above.
They aren’t foolish. They’re saving money. One human moderator costs more than deleting 10,000 accounts. One appeals agent costs more than issuing 10,000 automated rejections. Their big problem now is the bots and spammers. If the AI also deletes you, you are acceptable collateral damage regardless of how innocent you are. That’s the irresponsible corporate mindset.
Every time an account gets flagged, Facebook can either pay a human to review it or let an AI auto-delete it in a fraction of a second at near-zero cost and no liability. If 50,000 people are wrongly flagged, manual review would cost money. Auto-deleting them? Free and, FB believes, legally safe.
We need to understand that 50,000 people is nothing to a 3-billion-user platform. To you and me, it’s a filled-up Wrigley Field of kind of loving human beings. To Facebook, it’s 0.0016% of their base.
The accounts AI tends to delete are probably “low-value inventory”: older users, people who rarely post, folks who message family members, live outside high-ad revenue regions or don’t click ads. For Facebook, erasing them is no loss. And, AI is trained to be paranoid: better to delete 10,000 real people than let one “suspicious” account slip through.
In fact, word from some techies has it that Meta is deprioritizing Facebook. Their future is Instagram (young users + high ad revenue), WhatsApp (global communications monopoly), Meta AI products (future revenue goldmine). Facebook is now old, bloated, low engagement, declining in ad value. Deleting 50,000 Facebook users is like pruning a dying branch of a tree they no longer care about.
The only problem is that these are people. We are supposed to care about people. I care about people and that’s why I’ve written this. My heart goes out to the small business owners who lost money and the elderly who lost priceless photos. That’s why I am going to send this article to my Congressperson and Senators and I hope you will please contact your representatives as well.
This pruning is hurting people who thought they could trust a giant corporation. Meta reveals that it has no heart by doing this, it has no soul. A 60-year-old in Milwaukee who stores family photos and rarely clicks ads? AI deletes her. Fifty thousand like her? “Unprofitable.” Facebook isn’t a social network anymore, it’s an advertising and data collection machine. If you don’t generate revenue, and may cost them money, you’re expendable.
Facebook doesn’t seem to care about humans. It doesn’t seem to want to fix a broken AI (as per the BBC article). It doesn’t want to hear your appeals. It doesn’t want to save your memories, your small businesses, your family photos or your dignity. All it wants is to cut costs, collect data and pretend that “the algorithm” did the dirty work.
You didn’t break anything. You didn’t spam anyone. You didn’t cheat the system. You are not a robot and you can flip the bird like a pro to prove it. You just existed in a way that was inconvenient, and that’s enough for Meta to erase you without a second thought.
This is cruelty. If you think this is a glitch, think again, because this is the future of corporate AI and it’s already here. We should be very scared because there is no government body even trying to stand in their way; and they do not seem to have the character, integrity or empathy to do what a human being should. Yes, Meta, you failed to prove your humanity. I am deleting everything from you.
—
This Post is republished on Medium.
—
Photo credit: iStock
