澳大利亚研究人员发表在《Media International Australia》期刊上的最新研究表明，Google的AI概览功能在处理某些阴谋论相关搜索时存在缺陷1。研究对化学尾迹和15分钟城市两种阴谋论进行了对比分析，发现AI生成的摘要对这些虚假主张的反驳力度不足，部分情况下甚至将虚假声明呈现为合法辩论的一方1。其中，在15分钟城市相关的AI概览中，12.5%将虚假声明呈现为对立观点1。
对此，Google发言人回应称AI概览的绝大多数内容提供准确信息，并表示公司投入大量资源确保质量1。然而，德国慕尼黑法院已裁定Google对AI概览中的虚假声明承担法律责任1。研究人员呼吁Google等搜索引擎加强防护措施，防止阴谋论通过其平台得到传播和强化1。
Australian researchers have published findings revealing that Google's AI Overviews feature problematically handles searches related to conspiracy theories, sometimes reinforcing rather than effectively countering false claims.1 The study, published in Media International Australia in January 2026, examined how the AI-generated summaries addressed two conspiracy theories: chemtrails and the fifteen-minute city concept.1
The research identified a significant flaw in the system's handling of misinformation. In AI Overviews related to the fifteen-minute city conspiracy theory, 12.5% of cases presented false statements as a legitimate side of a debate.1 Rather than sufficiently refuting conspiracy claims, the summaries often failed to provide adequate counterarguments, inadvertently lending credibility to unfounded theories.1
Google has pushed back against the findings, with a company spokesperson stating that "the vast majority of AI Overviews provide accurate information" and asserting that the company invests substantial resources to ensure quality.1 However, the research arrives amid broader legal scrutiny of the feature. A Munich court in Germany has already ruled that Google bears responsibility for false statements appearing in AI Overviews.1 The Australian researchers are calling on Google and other search engines to strengthen safeguards to prevent the spread of conspiracy theories through their AI-powered search tools.1
评论
还没有评论，欢迎留下第一条。