As Americans prepare to vote in the US midterm elections, new University of Sydney research has found accusations of artificial intelligence (AI) use are creating an environment of distrust online that may affect how people engage with political news and elections globally.

The research found people are increasingly suggesting online posts and comments are AI-generated to discredit perspectives they disagree with. This growing trend risks exacerbating scepticism towards news on social media as a source of political information.

Led by Dr Timothy Koskie, researchers analysed 220,373 posts and comments from social media platform Reddit’s five largest news and political communities in the lead up to the 2024 US presidential election.

The study, published in Communication Research and Practice, found perceptions of AI use in election-related content can undermine trust in political information online and erode confidence in elections and democratic debate. While the findings are US-focused, they are relevant to elections worldwide.

“Many people now rely on online platforms to understand political issues and evaluate candidates for upcoming elections, and trust plays a vital role in that process,” said Dr Koskie, a member of the University of Sydney’s Centre for AI, Trust and Governance.

“Our research found widespread accusations on major social media platforms that AI is being used to manipulate political information, and these accusations were almost exclusively attached to contrasting political viewpoints.

“‘That’s AI’ has become a common way to dismiss information online, but what makes it so powerful is that it feeds into the misinformation and dishonesty that have grown with the proliferation of social media.”

Social media platforms, policymakers, technology companies and the wider community all have a responsibility to support healthy democratic debate online, Dr Koskie said.

“We should be critical of the information we encounter online, but the risk is legitimate concerns about AI are being weaponised to undermine democratic communication that is not AI-generated, potentially stifling robust political debate at a time when it is more important than ever,” he said.

Fake news, bots and algorithms

Using algorithmic analysis, the researchers identified recurring themes about AI use in political news and debate across Reddit discussions of the 2024 US election.

Concerns about AI misuse and interference – including unverifiable accusations of social media users being bots and platform algorithm manipulation – featured ubiquitously across the political discussions studied.

“References to bots and algorithms were frequently used to discredit information, particularly when users challenged claims or questioned the views of others,” Dr Koskie said.

“Posts and comments often accused other users of being politically motivated actors spreading misinformation to deepen distrust and inflate perceptions of ideological division.

“Just as Donald Trump’s repeated claims of ‘fake news’ encouraged distrust in journalists and media organisations who reported critically on him, we’re seeing a similar dynamic play out in online environments.”

Deepfakes and bots are a legitimate concern in news and elections worldwide, so any association with AI can label content untrustworthy even if it was written by a human, Dr Koskie said.

“When someone was labelled a bot, conversations lost momentum,” he said. “The accusation created uncertainty around who and what could be trusted, which influenced how people proceeded to engage with the political information they encountered.”

Another common theme was distrust of the algorithms used by social media platforms and the role they may play in shaping electoral outcomes. 

“Users frequently expressed concern about disinformation campaigns, the lack of transparency about how algorithms promote or suppress content, and the potential for platforms to influence public opinion for political or financial gain,” Dr Koskie said.

Strengthening trust in democracy in the AI era

With major elections taking place around the world, the researchers are calling for greater attention to the relationship between AI, trust and democracy.

“Our results confirm trust in political news and information in the age of AI is a defining challenge for democracy,” said Dr Koskie.

“Social media platforms have a responsibility to promote trustworthy and transparent information environments online, so people can engage meaningfully in elections and public debate.”

Political advertising and campaign messaging circulate through digital platforms, and voters need access to information they can evaluate with confidence when making decisions about candidates and policies.

“When unverifiable accusations of ‘bot armies’ and coordinated political manipulation are widespread, trust in the information environment weakens and it’s harder for credible information to gain traction,” said Dr Koskie.

“That erosion of trust makes it more difficult for people to navigate political debates and understand complex public issues.”

The study also highlights the importance of access to a range of independent news sources and perspectives.

“As AI use becomes more prevalent in online communication, it is important to ensure these tools support informed democratic participation,” Dr Koskie said.

Hero photo: Pedro Portal/Miami Herald/TNS/Sipa USA/AAP.