腾讯元宝回应生成拜年海报出现脏话

Core Viewpoint - The incident involving the Yuanbao App highlights potential issues in AI-generated content, specifically in maintaining appropriate language and context during multi-turn conversations [1] Group 1: Incident Overview - A lawyer in Xi'an experienced an offensive output from the Yuanbao App while generating New Year greeting images, despite not inputting any prohibited words [1] - The generated image changed from a positive greeting to an insulting phrase without any alteration in the visual elements [1] Group 2: Company Response - Tencent confirmed that the issue was due to an abnormal output from the model during multi-turn dialogue processing [1] - The Yuanbao team has taken immediate action to correct the problem and enhance user experience [1]