Meta's AI Chatbot Denies Trump Assassination Attempt: Company Attributes Error to 'Hallucinations'

Meta’s AI chatbot recently made a false statement denying the attempted assassination of former US President Donald Trump. The company has attributed this mistake to a malfunction in its chatbot technology. Joel Kaplan, Meta’s global head of policy, described these incorrect responses as “hallucinations” in a blog post published on Tuesday. 

Programming restrictions 

Meta’s original programming prevented the AI from discussing assassination

Initially, Meta’s AI was programmed not to respond to questions about the assassination attempt. However, this restriction was removed after it faced public inspection.

Despite the change, Kaplan acknowledged that “ in a small number of cases, Meta AI continued to provide incorrect answers, including sometimes asserting that event didn’t happen - which we are quickly working to address”. 

Censorship allegations 

Trump has accused Meta and Google of suppressing information     

The Problem of AI-generated misinformation extends beyond Meta. Google has also faced criticism for its Search autocomplete feature, which is accused of censoring results about the assassination attempt. In response, Trump urged his supporters in a Truth Social post to “go after Meta and Google”. He further accused both companies of censoring information related to his assassination attempt during a Pennsylvania rally.

Denial and apology

Meta and Google have denied any bias and issued apologies for the errors 

Meta and Google have rebut claims of bias in their information dissemination, attributing the issues to unintentional errors. A Meta representative apologized for the misinformation and announced that they are working on a fix to make sure more factual responses. Likewise, a Google spokesperson clarified that no manual intervention dominated their predictions and assured that developments are being executed.