Chatbots Struggle to Identify Fake Videos, Even OpenAI's Own Tools Are Deceived
AI video generation technology is developing rapidly, but mainstream AI chatbots still struggle to detect deepfake content. Research shows that when faced with fake videos generated by OpenAI's Sora, ChatGPT has an error rate of up to 92.5%, exposing serious shortcomings in AI's ability to judge the authenticity of content.