Edgar Alvarez* reports on the challenges facing Instagram to control the deluge of unpleasant bot messages that are just getting worse.
There isn’t a day when I don’t come across comments from bots on Instagram.
They’re all over the place.
But there’s nothing they love more than to spam high-profile pages with millions of followers.
No matter who they are, their posts are often the target of comments such as “We gonna ignore the fact that I’ve GOT A HUGE BOOTY?” or “DON’T LOOK at my STORY, if you don’t want to M A S T U R B A T E !”
Behind these, are fake accounts featuring pictures and videos of naked and half-naked women, whose primary goal is to get people to sign up for shady porn sites.
This has become a serious problem for Instagram, one that seems to be getting worse by the day and that the company needs to get a handle on, before it gets more out of control.
On April 21, two days after we first reported on how these bots were invading Instagram, celebrity and social media influencer Chrissy Teigen tweeted about the issue to her more than 11 million followers.
“Honestly Instagram needs to handle this shit,” she said.
It was a reply to a tweet from someone who had taken a screenshot of the comments section in one of Teigen’s Instagram posts, which showed it being flooded by spam comments like the examples given above.
“I don’t know how you cope,” the tweet to Teigen said, referencing comments including “don’t look at my story, if you don’t want to M A S T U R B A T E !” and “we gonna ignore the fact that i’ve GOT A HUGE BOOTY,” which are two of the most common variations of this type of spam.
What makes these sketchy comments thrive is that they tend to garner hundreds or thousands of likes instantly, sending them to the top of the comments section on posts from celebrities and other popular accounts.
Some of those likes come from bot networks, others from regular people who just happen to find them funny.
Sean Spielberg, co-founder & CEO of Instascreener, an analytics agency that focuses on influencer marketing, says it is quite easy to write a script that continually pings an Instagram account and checks for new posts.
“One of these [spam] networks could check LeBron’s profile every second and, as soon as a new post goes up, add a comment immediately and have other accounts like [it] comment,” he said.
“The network could also continually leave multiple comments from multiple accounts to make sure theirs are the most recent.”
It’s unclear how many of these spam profiles are out there, but a recent report from Instascreener claims there are over 150 million fake accounts on Instagram.
Facebook told Engadget it couldn’t disclose specific numbers, but the company says that every day it blocks “millions” of attempts to create fake accounts during the sign-up process on Instagram.
The Facebook-owned social network has over 1 billion monthly active users, so even if only one percent of those accounts were bots, that would still amount to ten million fake accounts.
To give you an idea of the extent to which Facebook has to combat bad actors on its family of apps, in May the company revealed it took down 2.19 billion fake accounts during the first quarter of 2019 alone.
That’s quite a staggering figure when you consider that Facebook has 2.4 billion monthly active users.
According to a report by Tenable, a cybersecurity research firm, many of these “porn bots” are coming up with new techniques to avoid being detected by Instagram’s security systems.
This may come as a surprise, but that includes using lines from Game of Thrones in their image captions — which tricks Instagram’s systems into thinking they’re being written by a real person, not a bot.
Facebook says the challenge with this is that it needs to ensure that the methods it puts in place to fight spam, as well as other inauthentic engagement, doesn’t end up affecting real people.
For instance, Facebook said, it’s difficult to train technology to tell the difference between a Game of Thrones comment coming from a normal user like you and another from an account that might be fake.
But these porn bots aren’t just using pop culture reference to try to make it harder for Instagram to distinguish between them and an authentic user: They’re also leaving comments with weird spacing between letters and grammatical errors.
The reason spammers might format their comments this way is because, in 2016, Instagram started letting people filter words out of post comments.
So, while it may be easy for someone like Teigen to block the word “masturbation” from showing up in her posts, “M A S T U R B A T I O N!” may require a little more work on the user’s part.
Facebook says it is fully aware of spam/porn bots on Instagram, noting that it is investing more in research to better understand how these bad actors are evading its systems.
And, more importantly perhaps, the company says it is working to build tools to get rid of these bots more quickly and efficiently.
“Nobody likes receiving spammy follows, likes and comments,” a Facebook spokesperson said to Engadget. “It’s really important to us that the interactions people have on Instagram are genuine, and we’re working hard to keep the community free from spammy behavior.
“Bad actors continue to do everything they can to get around the measures we’re putting in place, and it’s our job to stay one step ahead.”
Satnam Narang, the author of Tenable’s report, says the porn bots are now so sophisticated that not only can they like your pictures or comment on them, but they can even slide into your direct messages.
Regardless of how they try to interact with you, though, the intention is always the same: to peddle dubious dating and webcam sites for adults.
“To its credit, Instagram has worked to try to thwart the efforts of the operators of these porn bot accounts,” he said. “But, as you can imagine, it is a cat-and-mouse game.”
Facebook told Engadget it is committed to figuring out the motives and tactics of these bad actors on Instagram, and that its abuse-fighting team is constantly updating its automated and manual systems to help detect any suspicious activity on the app.
One of the ways it’s doing that, Facebook said, is by using machine learning to examine thousands of attributes from accounts and focus on behaviour that’s difficult for spammers to fake, like their connection to real people on Instagram.
In addition, the company says it is investing heavily in tackling inauthentic engagement, which makes it easier to spot when an account is using a third-party service to generate fake likes, comments and followers.
Given the massive scale of Instagram, this problem isn’t going to be fixed overnight, however.
And while Facebook is adamant that it is dedicating plenty of resources to combat spam/porn bots on Instagram, the company said something similar months ago and not much has changed.
If anything, these comments are more prominent now than they were in April, when we first reported on them.
Still, Facebook says it is investing in this area for the long term, and the hope is that one day soon you won’t come across any of these accounts on the app.
Until then, don’t be surprised when you check out a celebrity’s post and keep seeing comments from random accounts telling you to stop ignoring their huge booty.
* Edgar Alvarez is a Senior Editor for Engadget. He tweets at @abcdedgar.
This article first appeared on www.engadget.com
