Meet ‘DebunkBot’ can AI truly combat conspiracy theories?

In the digital age, where misinformation spreads rapidly, artificial intelligence has emerged as a potential solution for countering conspiracy theories.

One such example is “DebunkBot“, an AI chatbot designed to engage users who believe in conspiracy theories. A recent study published in the in Science demonstrated impressive results: After brief conversations with the bot, participants’ belief in conspiracy theories decreased by 20 percent, with approximately a quarter of them abandoning those beliefs entirely. Moreover, these effects persisted even two months later, suggesting that AI could provide a long-term solution in challenging misinformation.

As Gordon Pennycook, one of the study’s authors, remarked, “The work does overturn a lot of how we thought about conspiracies,” challenging the long-held belief that conspiracy theories cannot be debunked through facts and logic alone due to cognitive dissonance — the discomfort people feel when confronted with information that contradicts their deeply held beliefs. DebunkBot appears to bypass this barrier.

However, the success of DebunkBot was largely due to its ability to provide tailored, fact-based responses to users’ specific concerns, rather than relying on generic debunking strategies. This personalized engagement allowed the bot to address each user’s unique beliefs, overcoming cognitive biases such as confirmation bias, which often fuel conspiracy theories.

While these results are promising, they also raise broader questions: Can AI alone truly combat conspiracy theories in a world increasingly characterized by distrust of traditional institutions and information sources? Brendan Nyhan, a misinformation researcher at Dartmouth College, highlights a critical concern: “You can imagine a world where AI information is seen the way mainstream media is seen” — with widespread skepticism and distrust. If AI becomes viewed as just another tool of elites or tech companies, it may fail to gain the trust of those it seeks to help, particularly in a context where many conspiracy believers already harbor deep distrust of traditional media and institutions.

Terry Flew, a professor of digital communication and culture in the Department of Media and Communication at the University of Sydney, argues that the erosion of trust in news media and political elites has been a significant factor in the rise of populism and conspiracy theories. He emphasizes that misinformation thrives in environments where distrust is prevalent. In this context, AI interventions like DebunkBot must be part of a larger effort to rebuild trust, rather than just providing factual corrections.

Flew highlights a global trust crisis in political, social and media institutions, and outlines three interconnected levels of trust: macro (societal), meso (institutional) and micro (interpersonal). This framework is crucial when considering AI’s role in combating conspiracy theories because AI functions at the intersection of these levels — providing factual corrections (micro), working within trusted platforms (meso) and influencing broader societal beliefs (macro). Flew concludes that AI interventions must address this broader trust deficit in society to be truly effective.

Similarly, in The Psychology of Conspiracy Theories, social psychology professor Karen M. Douglas explains that when people feel a lack of agency, they find the simplified explanations offered by conspiracy theories appealing. This psychological need complicates efforts to debunk these beliefs with facts alone, as cognitive biases such as confirmation bias further entrench conspiracy thinking. Both Flew and Douglas highlight that conspiracy theories thrive when individuals feel disconnected from institutional elites and media gatekeepers.

The challenge for AI interventions, as MIT professor David Rand notes, is ensuring they resonate with users while remaining transparent, neutral and grounded in accurate information. This is especially crucial in an era where trust in traditional media has diminished, and social media has become fertile ground for conspiracy theories.

Research on social trust indicates that people are more likely to trust institutions—and by extension, AI systems—when they are perceived as transparent, accountable and working in the public interest. Conversely, if institutions or AI tools are viewed as opaque or manipulative, trust is eroded. For tools like DebunkBot to remain effective, they must consistently demonstrate neutrality and fairness. Policymakers, developers and technologists must work together to ensure AI systems are seen as trustworthy participants in the fight against misinformation.

While DebunkBot shows promise, challenges remain. AI interventions need to strike a balance between being persuasive and maintaining user trust. Integrating AI into everyday platforms, such as social media or health care, where conspiracy theories often circulate, could improve effectiveness. For instance, AI could be used in doctor’s offices to debunk vaccine myths or in online forums where misinformation spreads rapidly.

The long-term success of AI in combating conspiracy theories will require more than technological innovation. Trust, psychological factors, and public perception will all play crucial roles in determining whether AI can make a lasting impact. As Flew emphasizes, rebuilding trust in media, institutions and the broader information ecosystem is essential for any solution to succeed. AI tools like DebunkBot can play a critical role in providing personalized, fact-based interventions, but they must be part of a larger strategy involving transparent communication, collaboration between policymakers and developers, and efforts to address the broader societal trust deficit.

Pari Esfandiari is the co-founder and president at the Global TechnoPolitics Forum, a member of the at-large advisory committee at ICANN representing the European region, and a member of APCO Worldwide’s advisory board.