Many of the Twitter activists hoping to #ReopenAmerica may not have been under COVID-19 lockdowns after all.
A recent Carnegie Mellon report found that almost half of all the tweets using this #ReopenAmerica came from bots, leading me to wonder how these bots continue to infiltrate our social media in such an unstoppable manner. So, I decided to make my own.
Originally, my goal was to simply understand the essence of these bots and how they propagate information so quickly. Quickly, I found that all it takes is a computer, a few Youtube videos, and some basic coding knowledge to create a potentially potent bot. I first coded a program to scrape all of the articles featured on The Politic’s website. After running the program, I was left with a text file containing links to every single article on the webpage.
For most internet bots, data scraping is vital to their performance on social media platforms like Twitter. The more articles and media they can collect, the easier it is to continuously post data through automated accounts. In fact, it only took me a few different lines of code to convert my original scraping program for The Politic to one that pulls all the top articles from the New York Times, The Washington Post, Breitbart, or any other online news publication.
In large-scale bot operations, this process is likely automated and optimized to fit the needs of the organization contracting them. If a political group wanted to promote pro-life media, they may employ artificial intelligence to generate a list of internet articles that support their cause. As data scraping techniques become more advanced, any article published on the web could suddenly be posted or retweeted by a bot halfway across the world.
Internet bots are increasingly being used to promote various stories—some true and some not so much—from media sources that align with the agenda of those controlling the bots. This supplies the ammo that political actors need to generate Twitter accounts that can share massive amounts of information to the world, including one case in which Tinder bots were used to promote the agenda of the Labour Party.
Not only do these operations scrape from news media sites, but they can also pull from the very social media platforms they plan to influence. Most internet bots today can scrape analytics about certain hashtags or even create their own trending tags. A 2018 investigation by the BBC found that Saudi Arabian companies were offering to use bots to create trending hashtags for as little as US$200. Taking a look back at #reopenamerica, it’s easy to wonder if other popular movements were simply a ruse.
But once all this data is scraped from the web, where does it all go? Once I collected all of the URLs from The Politic’s website, it was now time to take it to Twitter. I created a new Twitter account under the handle @ThePoliticBot and applied for a Twitter Developer account to interact with my profile via code. Within five minutes, I was approved and set up with everything I needed to interface with the account.
Once I was approved, the next step was to write a script that would publish a tweet at a given time interval. To make this bot work, I decided to publish a random article every hour along with a caption to make the posts seem a bit more real. With just 23 lines of code and a small list of potential caption options, the bot went live and was ready to go.
In just an hour, I had set up an automated Twitter account capable of tweeting articles, media, or almost anything else you could think of. Although this bot was created for relatively tame purposes, it made me question the protections stopping someone from creating an account for nefarious reasons. With just a different data scraping technique and a different motive, this innocent article posting bot could be turned into a propaganda machine, posting and retweeting pieces of stories and misinformation that could spread like a wildfire.
While the process involved creating an account and applying for developer access, many bot operators bypass these steps entirely. Instead, they retrieve the login information of inactive accounts through account information scraped from hacks of other social media platforms, making it seem like a real human being has simply gotten back on Twitter. A prime example of this is @gearsussx, a Twitter account that President Trump retweeted in 2016. The Twitter page was created in March of 2013 with little to no activity for the majority of its existence. However, from March 20 to March 30 of 2016, the account posted an astonishing 1,150 tweets, most of which being pro-Trump media. The account was removed after multiple publications exposed suspicions of bot activity.
The ease of creating these automated tools is what has stunted social media platforms such as Twitter from taking any meaningful steps in fighting against bots.
As bots become more and more advanced, it becomes harder to distinguish them from human accounts, as programmers are becoming more creative with the way they disguise the persona behind these automated tweets. The odds even suggest that most of us have interacted with a bot on Twitter before without even noticing. The Pew Research Center found that across all the posts on Twitter that link to a website, about two in three come from a user that shows signs of bot activity. With such an overwhelming presence on these social media platforms, it’s understandable why political groups invest so much money in digital media marketing where the creation of bots can help spark support for a certain agenda.
Even when Twitter is able to identify the spread of misinformation using emerging tools such as AI, the company faces multiple roadblocks in its attempts to eliminate these false streams of stories. As platforms that are extensions of the tenets of free speech, censoring their platform is often heavily criticized by the public as a method of controlling the information that their users see. Although there is no denying that social media platforms have often been complicit in the censorship of various information, it is important to note that regardless of the beliefs of its users, social media companies have the obligation to call out and remove misinformation that is touted as fact.
Although Twitter has released multiple statements detailing the methodology they are using to tackle misinformation, it won’t be getting easier for them anytime soon. On May 28, President Donald Trump announced the signing of an Executive order aimed at companies including Twitter. The order targets social media platforms for censoring certain political content by claiming that these companies do not act in good faith. They state that by censoring information, companies violate Section #230 of the 1996 Communications Decency Act, an ordinance that states that media companies are not considered publishers of the content posted by users on their platform.
This order not only puts social media platforms at risk for the infiltration of misinformation, but it actually makes it harder for these companies to remove malicious automated accounts from their platform. If companies were to eliminate accounts that promoted certain beliefs with misinformation, they may come under scrutiny for violation of the Trump administration’s interpretation of the Communications Decency Act.
With bots becoming a ubiquitous part of our social media experience, it is important that we—as well as politicians—be mindful of the struggles of handling these covert political actors and call out misinformation, regardless of whether a human presents it or not.