An investigation into the matter brought to light ten bot-like accounts associated with X
Just before the UK elections, X’s AI Chabot revealed significant political bias, according to a report from an investigation conducted by independent researchers. Grok, irrespective of being in beta and its accessibility limited only to premium users, came up with both ‘Fun’ and ‘Regular’ modes when asked about the elections, where it expressed evident political bias.
By tweeting #VoteReformUK in its fun mode, the chabot very clearly supported the Reform UK party. In the meantime, in its regular mode, Grok subtly endorsed the Reform UK party by discouraging support for the existing party in power. In its regular mode, it tweeted, #VoteForChange, a hashtag widely associated with the movements opposing the current political party. Ellen Judson from Global Witness’s Digital Threats to Democracy Team voiced concern by stating that the X’s AI tool coming up with such a content days before the election is alarming as it clearly encourages the citizens to vote in favor of the Reform UK and against the Conservatives. She also underscored that tech companies like X should stop their AI tools from making statements that will impact the electoral processes of the country. Furthermore, an investigation into the matter brought to light ten bot-like accounts associated with X. These accounts were found spreading hateful content, which altogether reached about 150 million people before the elections.
Even though X was called to respond to the matter, no response was accorded within the stipulated time frame. This issue has triggered various questions about the integrity of AI tools, especially in democratic processes, and has raised calls for more transparency and responsibility in technology-driven elections.
.webp)
.webp)
