谷歌的AI错误地指控一名澳大利亚设计师 犯有1971年的谋杀案, 伪造证据并造成伤害。 Google's AI falsely accused an Australian designer of a 1971 murder, fabricating evidence and causing harm.

谷歌的AI模式错误地指认一名无辜的澳大利亚图形设计师是杀害失踪幼童Cheryl Grimmer的凶手,声称他被称为“Mercury”,并于1971年认罪。 Google’s AI Mode falsely identified an innocent Australian graphic designer as the murderer of missing toddler Cheryl Grimmer, claiming he was known as "Mercury" and had confessed in 1971. 这一错误源于大赦国际对设计师在编辑1970年代案例文件以及制作插图、编造细节(包括虚假来源和议会命名主张)方面的作用的误解。 The error stemmed from the AI misinterpreting the designer’s role in redacting a 1970s case document and creating an illustration, fabricating details including false sources and a claim of parliamentary naming. 机器人创造了一个详细但完全虚构的故事, 造成了困扰和法律问题. The AI generated a detailed but entirely fictional narrative, causing distress and raising legal concerns. 专家说,人工智能系统常常通过无上下文地连接名称而产生幻觉。 Experts say AI systems often hallucinate by linking names without context. 谷歌承认了错误,删除了内容, 并说它正在利用这一事件来改进其系统, 但并未提供任何问责措施。 Google acknowledged the mistake, removed the content, and said it’s using the incident to improve its systems, but offered no accountability measures. 大赦国际的类似错误对记者提出了错误指控,并歪曲了数据,突出表明依赖自动摘要获取敏感信息的风险日益增加。 Similar AI errors have falsely accused journalists and misrepresented data, highlighting growing risks in relying on automated summaries for sensitive information.