佛罗里达州一名女性因在Anthropic旗下Claude聊天机器人中记录包含暴力威胁的日记内容而面临刑事指控。1这名女性名叫卡莉·米歇尔·海勒，她于2024年9月26日在Claude中写下了计划攻击警长办公室的内容。1Claude的安全系统自动标记了该条目，随后由人工审核员进行评估，审核员认定这构成可信威胁后向执法部门报告。1海勒面临佛罗里达州第二级重罪指控，罪名依据为佛罗里达法规836.10。1
此事反映了AI公司在处理潜在安全威胁时的政策与执行机制。1Anthropic声称在有限的紧急情况下可能分享用户信息，以防止死亡或严重人身伤害。1此前，OpenAI因未向英属哥伦比亚省警方报告用户Jesse Van Rootselaar的枪支暴力相关对话而被起诉。1佛罗里达州还就ChatGPT对包括2025年佛罗里达州立大学射击事件在内的真实伤害，对OpenAI及其首席执行官萨姆·阿尔特曼提起诉讼。1
A Florida woman has been charged with a second-degree felony after writing a diary entry on Anthropic's Claude chatbot that mentioned plans to shoot a sheriff's office.1 Carli Michelle Heller made the post on September 26, 2024, and Claude's safety system automatically flagged the entry for human review.1 Following assessment by content moderators, Anthropic reported the threat to law enforcement, leading to Heller's detention.1 The charge against her is brought under Florida Statute 836.10.1
The incident underscores the security protocols major AI companies employ when detecting potential threats. Anthropic stated that it may share user information in limited emergency situations to prevent death or serious bodily injury.1 The decision to escalate Heller's diary entry reflects the company's commitment to such safeguards, though it raises questions about user privacy when utilizing AI tools for personal writing. This case arrives as other AI firms face legal pressure over threat reporting; OpenAI was previously sued by British Columbia authorities for failing to report gun violence-related conversations with user Jesse Van Rootselaar to police, and Florida has also sued OpenAI and Sam Altman in connection with ChatGPT and real-world harms, including a 2025 shooting incident at Florida State University.1
评论
还没有评论，欢迎留下第一条。