X Plagued by Bots Amplifying Hate Speech and Disinformation, Global Witness Investigation Reveals
A recent investigation by Global Witness has uncovered a network of approximately 45 bot-like accounts on X (formerly Twitter) that have amassed over 4 billion views while disseminating harmful content, including racist and sexualized abuse, conspiracy theories, and climate disinformation. These accounts, identified through their activity leading up to the UK general election, generated an estimated 610,000 posts between May 22 and July 22, 2024. While the investigation sampled accounts supporting the four largest political parties in the UK, a disproportionate number of the bot-like accounts expressed support for Reform UK.
The investigation’s findings highlight the ease with which automated accounts can manipulate online discourse and spread harmful narratives. After the UK election, these accounts shifted their focus to international events, including anti-immigrant protests in Ireland, the alleged assassination attempt against Donald Trump, and speculation regarding Joe Biden’s potential withdrawal from the 2024 US presidential race. Their responses to these events were often laced with racism, gendered disinformation, and unfounded conspiracy theories, demonstrating a clear pattern of exploiting sensitive topics to sow discord and amplify divisive rhetoric.
Ava Lee, Campaign Leader of Digital Threats at Global Witness, expressed alarm at the proliferation of these accounts and their potential impact on democratic processes. She emphasized the crucial need for access to accurate and reliable information, particularly in the lead-up to major elections like the upcoming US presidential race. Lee called upon X to bolster its moderation efforts and enforce its existing policies against inauthentic activity, urging the platform to take greater responsibility in protecting democratic discourse from manipulation and interference.
While there is no evidence linking UK political parties to the funding, use, or promotion of these bot-like accounts, their distribution across party lines is noteworthy. Of the identified accounts, 28 posted in support of Reform UK, 14 supported the Labour Party, and a smaller number aligned with the Liberal Democrats and Conservatives. The investigation revealed a distinct partisan divide in post-election activity, with 14 of the Reform UK-supporting accounts actively promoting climate change denial through hashtags like #ClimateScam.
The investigation further revealed the adaptability of these bot-like accounts, swiftly pivoting to exploit emerging news and trending topics. In the wake of the alleged assassination attempt against Donald Trump, several accounts previously supporting Reform UK propagated conspiracy theories suggesting Joe Biden’s involvement, while some Labour-supporting accounts questioned the authenticity of the incident, suggesting it was staged by Trump himself. Similarly, following speculation about Biden potentially stepping down from the presidential race, numerous bot-like accounts intensified their focus on Kamala Harris, sharing racist, sexist, and disparaging comments, often employing well-known tropes of gendered disinformation.
The investigation also documented the participation of these accounts in online discussions surrounding anti-immigrant protests in Dublin, Ireland. These accounts disseminated conspiracy theories and used xenophobic hashtags such as #IrelandbelongstotheIrish, further fueling divisive narratives. The consistent promotion of climate change denial through hashtags like #ClimateScam and #geoengineering underscores the coordinated nature of these accounts and their intent to manipulate public opinion on critical issues. Global Witness contacted X for comment on these findings but received no response.
Methodology of the Global Witness Investigation:
The Global Witness investigation employed a rigorous methodology to identify and analyze bot-like activity on X. The research focused on accounts promoting the four major UK political parties – Conservative, Labour, Liberal Democrat, and Reform UK – by sampling posts containing relevant hashtags like #voteconservative, #votelabour, etc., during the six-week period leading up to the UK general election.
The investigation relied on identifying specific "red flags" that suggested inauthentic, automated, or disproportionately high participation in political discussions. These red flags included excessive posting frequency (over 60 times a day on average, and in some cases over 144 times), posting in multiple languages, a high proportion of reposts (over 90%), a low volume of original content (less than 10%), posting duplicate content, recently created accounts (within the last 50 days), suspicious account names and handles, and consistent engagement metrics (likes and reposts) suggesting potential coordination. Accounts exhibiting these red flags were manually reviewed to assess for any evidence of authentic human identity or personal interactions.
The investigation also classified accounts based on their political affiliation by analyzing hashtag usage, account bios, profile and cover photos, posts, and shared media content. While the study identified these potential bot accounts and their activity, it does not offer definitive proof they are part of coordinated bot networks or troll farms, recognizing the inherent difficulty in definitively confirming such connections. Furthermore, the investigation emphasizes a lack of evidence suggesting that any UK political party is involved in financing, utilizing, or promoting these bot accounts for election campaigning. The figures presented in the report are estimates and may vary due to account or post deletions. The research was supported by Information Tracer.
