普林斯顿大学研究员麦克斯·施普林格通过结合人工智能大语言模型和公开选举数据，成功识别了乔治亚州98.9%的提前投票选票，揭示了该州投票系统中的严重隐私风险。1施普林格仅用20美元的AI订阅成本，在几小时内就建立了分析管道，无需亲临乔治亚州便恢复了1.52万张选票的投票顺序，涵盖114个县中的人工投票。1在选民较少的县份，这一漏洞尤为明显——以赫德县为例，该县650张提前投票中的大多数都能被AI与特定选民匹配。1
这一发现促使乔治亚州选举委员会在中期选举提前投票开始前不到两周召开紧急会议。1州务卿拉芬斯珀格已下令在选举后删除选票ID号以补救此问题，1但这一措施与州法律对投票保密和选举记录透明度的要求存在冲突。1州秘书处代表透露，该部门已向立法机构提出资金帮助请求长达三年但未获批准。1与此同时，其他州要么已对相关软件进行了修补，要么限制了易受攻击的投票数据的公开检查。1
Researchers have identified a significant privacy vulnerability in Georgia's voting system, revealing that artificial intelligence can be used to determine how individual voters cast their ballots. Princeton University postdoctoral researcher Max Springer successfully reconstructed the voting patterns of 15,200 early votes—representing 98.9 percent of advance ballots across 114 Georgia counties—by combining a $20 AI subscription with publicly available election records.1 Working without ever visiting Georgia, Springer built his analysis pipeline in just a few hours.1 The breach is particularly acute in less populous counties; in Heard County, for instance, where 650 early votes were cast, the AI was able to match the majority of individual ballots to specific voters.1
The discovery has prompted emergency action by Georgia's election officials. The state election board convened an urgent meeting just weeks before early voting began to address the vulnerability.1 Secretary of State Brad Raffensperger subsequently ordered the deletion of ballot identification numbers following Election Day to mitigate the exposure.1 However, this remedy creates a legal conflict within Georgia, as state law contains competing requirements between protecting voting secrecy and maintaining transparency in election records.1 State officials acknowledged that they have sought funding from the legislature for three years to resolve such issues without success.1 Meanwhile, other states have either patched vulnerable software or restricted public access to sensitive voting data.1
评论
还没有评论，欢迎留下第一条。