#misinformation

'An eight-minute conversation with a GPT-4 based chatbot reduced belief in conspiracy theories for many participants, with effects lasting at least two months'

'Sora is being used to create believable deepfakes that scammers exploit, highlighting a widening digital trust crisis.'

'Deepfake videos have moved from niche stunts to mainstream news risks, forcing media and platforms to rethink verification and regulation.'

Volunteers are labeling and reviewing suspected 'AI slop' on Wikipedia to protect the site's trustworthiness and prevent AI-generated inaccuracies from spreading.

AI chatbots like ChatGPT have been criticized for being overly agreeable, often affirming users' statements whether true or false. This article explores why this happens, the risks involved, and how developers and users can work to improve chatbot reliability.