尽管人工智能（AI）会议记录工具承诺能自动生成会议纪要，但部分专业人士对其引发的数据隐私和安全隐患表示强烈担忧 [1]。这些工具会将会议中的一切言论转化为数据进行存储或用于模型训练，可能导致机密人员信息、商业策略、贸易秘密甚至可能被视为有罪供述的言论落入他人手中 [1]。
人力资源培训与认证提供商 HRCI 的首席执行官 Amy Dufrane 指出：“组织在使用 AI 记录工具方面存在巨大风险。”她进一步表示，“我认为公司根本不应该使用此类工具”。亚特兰大律师行 Baker Donelson 的资深企业律师 Justin Daniels 也表达了类似的疑虑，称“使用 AI 会议记录工具的人并不总是清楚数据去了哪里”。
隐私倡导者担心，这些工具的开发商在未经同意的情况下创建了声音指纹（voiceprints）。伊利诺伊州法律规定，声音指纹被视为与指纹相似的生物识别标识符，受该州《生物信息隐私法》管辖；该法案要求在 AI 记录工具或其他代理收集声音指纹前必须提供书面通知并获得知情同意 。此外，一些科技公司会转售其开发工具的录音数据，或利用机密会议记录和转录内容来训练其人工智能模型 。
近期发生的一起案件凸显了此类风险：今年二月，一名纽约联邦法官命令刑事被告向检察官提供其为律师制作的文件，原因是这些文件此前已与第三方共享，该第三方即为 Anthropic 旗下的 Claude。
While artificial intelligence tools promise effortless meeting summaries by converting everything said into data [1], some professionals warn of significant security and privacy risks associated with their use. Amy Dufrane, chief executive of human resources training provider HRCI, stated that there are huge organizational risks to AI notetakers and does not believe companies should utilize them at all ["There are huge risks to the organization on AI notetakers," said Amy Dufrane... "I don't think companies should use it at all."] [1].
Concerns center on how these tools handle sensitive information. Confidential personnel data, corporate strategies, trade secrets, and remarks that could later be viewed as incriminating may end up in the wrong hands ["Confidential personnel information, corporate strategies, trade secrets..."] [1]. Privacy advocates further worry that companies behind these AI notetakers are creating voiceprints without consent [1]. In Illinois, where voiceprints are considered biometric identifiers similar to fingerprints and covered under the state's Biometric Information Privacy Act requiring written notice and informed consent before collection by an agent like an AI notetaker ["In Illinois..."], this lack of transparency is particularly problematic.
The issue extends beyond simple data storage; some tech companies resell data from their tools or use confidential meeting transcripts to train their AI models [1]. Justin Daniels, a corporate attorney at Baker Donelson based in Atlanta, noted that users often do not know where the data goes ["People who use AI notetakers..."] [1]. Legal precedents highlight these dangers: a New York federal judge ordered a criminal defendant in February to provide prosecutors with documents created for his lawyers because they had already been shared with Anthropic's Claude as a third party ["a New York federal judge in February ordered..."] [1].