I Paid an AI to Get Me Instagram Followers. It Didn't Go Well

The idea of Instagram automation is simple: you pay a computer to find hashtags, content and people that match your brand, and interact with them to drive up engagement on your account. These services aren’t not illegal, but they do violate Instagram’s platform policy and can get you banned. Still, they’ve been used by businesses, political consultants and even (allegedly) American Idol contestant Clay Aiken and Real Housewives reality star Sonja Morgan.

And they’re evolving. Using artificial intelligence, today’s automation services have been tweaked to better emulate human behaviour and avoid the Instagram algorithm that spots suspicious activity. More premium options can recognise the content of images, so the comment “you” leave on someone’s selfie will be different from one on, say, a photo of a delicious sourdough pizza. Some bots even analyse the results of their efforts on the fly, constantly adjusting which accounts to interact with. 

This is where my downfall began.

I set out to compare a number of competing automation services to gain insight into the industry. The first one I pick offered a money-back guarantee if I didn’t see results in two weeks.  A couple of days pass. A week. Then ten days. Nothing. I request a refund. With my $49 returned to my credit card as promised, I decide to pull out the big guns. The competitor I settle for is more expensive – roughly $60 monthly – and can watch people’s stories, target trending hashtags and even respond to any comments people leave on your stories or posts. Sold.

Now, giving up control of your personal social media account to an AI feels like it could lead to potentially awkward situations. As I hand over my Instagram password to a bot, I realise that I have to set some boundaries. I don’t want this thing anywhere near my boss or coworkers – really, nowhere near anyone I actually know. 

But the AI has just as many settings as it does features. After an hour’s exploration of the complicated interface, written in what I suspect to be Google Translate English, I find a “blacklist” setting. Anything you enter here – hashtags, locations or usernames – will be skipped. I produce a list of names: potential employers, friends I’ve fallen out with, ex-boyfriends. I feed the list to my bot, confident it won’t be able to cause any trouble.

Just like last time, the following couple of weeks are uneventful. During the third week, however, strange things start to happen. With increasing frequency, I find myself getting logged out of Instagram due to “suspicious activity”. From my new automation provider’s website, I pay another $10 (per month) for a VPN add-on supposed to remedy the problem. It works. The logouts stop. Then I kind of just forget about it. With nothing unusual going on whenever I open Instagram, another three weeks or so pass with me not thinking about my bot at all (it’s called automation, after all).

One particularly boozy Friday night, I end up in bed with my boyfriend, gossiping about people on Instagram. One of them is a guy my ex is currently seeing. On Sunday morning, I notice that same guy appearing in my feed. Disturbed, I realise I must’ve accidentally followed him when looking him up. Worse, I’ve also somehow liked his latest post, a picture of him having dinner with a friend. 

Faster than anything I’ve ever done, I remove the like and unfollow him. But with two days having passed since the night of the crime, there’s no way he hasn’t noticed. I debate whether I should text my ex and explain the mistake, but decide against it. My ex and I don’t talk anymore and I’d rather spare myself more embarrassment; the damage is done. Or so I thought.

Now, it might seem strange that I’d so easily internalise the idea that I was the perpetrator here, despite having no recollection of committing the crime. Truth is, this wouldn’t be the first time I embarrassed myself online. Like that time I posted a selfie making fun of straight people on the Instagram of a multinational finance company I worked for, or the time I was complaining about someone I was seeing to a friend, only to look down on my phone and see my monologue saved as a voice recording and sent to that very guy. 

These past offences are well-known in my circle of friends, who all agree: It would be best for everyone if I weren’t allowed to operate anything more advanced than a burner phone. So when disaster strikes, I immediately bypass denial, anger, and the remaining stages of grief, jumping straight to acceptance: of course I did it.

Things go from bad to worse when one morning, I get a text from my ex. The tone isn’t very nice. He’s asking me to stop following and deep liking “literally everyone I come in contact with on social media”. Friends and coworkers being “disturbed” is enough; moving on to the guy he’s seeing and his friends truly crosses the line. My ex lets me know “everyone” is comparing me to Luca Magnotta, infamous killer of people and cats, known from Netflix's Don't F**k With Cats.

When I try to explain that following his boyfriend and liking his content was an honest mistake, he doesn’t buy it. After a short, unpleasant exchange of texts, I decide to block him. But blocked or not, my ex’s text has me brooding over “the incident”. Yes, I’m clumsy. But this clumsy? And something about the whole “friends and coworkers” thing bothers me. I tell my friends the whole thing is making me feel a little crazy.

It’s only when a middle-aged woman sends me a DM that it strikes me. She wants to know if we’ve ever met. When I tell her I don’t think so, she asks why I followed her. The realisation is uplifting and at the same time terrifying. My bot has gone on a rampage, liking, following and commenting on accounts like a crazed stalker. 

My stomach sinks as I think of all the people “I” may have reached out to and what message they’d think I was trying to send. My ex’s boyfriend: “I see you.” That friend I had a fight with: “I’m ready to forgive” (I’m not). An acquaintance who’s very clearly not single: “Dump him.” Clearly, the blacklist feature didn’t work as expected. My best guess is that it may have skipped direct interactions with the specified accounts, but that it can still target accounts connected to theirs. 

If you run a business account, that might be fine. It’s not fine for me. I don’t know what exactly my bot has been doing the past two months. More importantly, I have no idea of with whom. I set out to review today’s best automation services, but after the second one (and $140 poorer), I’m out. I log in to my bot’s interface one last time, pause all of its activities, and cancel my subscription. I close my browser window. Then I close my eyes and pray I’ll never find out what else my bot has done.