The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
A Swiss study found that large language models like ChatGPT are significantly more persuasive than humans in online debates, especially when given personal data about opponents. Researchers warn this capability could be misused for disinformation, impersonation, and manipulating public opinion, posing future risks to communities and democratic processes.[AI generated]
Why's our monitor labelling this an incident or hazard?
The article explicitly involves AI systems (large language models) and discusses their use in persuasion and influence in online debates. It highlights the potential for these AI systems to be used maliciously to spread misinformation and impersonate individuals, which could plausibly lead to harms such as violations of rights, harm to communities, and disruption of democratic processes. Since the harms are not reported as having occurred yet but are clearly plausible and credible risks, this event qualifies as an AI Hazard rather than an AI Incident. The article does not describe a realized harm but warns about credible future risks associated with AI use in persuasion and misinformation.[AI generated]