巴西最高选举法院周二发布新规，对政治广告中的人工智能生成内容制定了明确的监管标准。1法院将深度伪造定义为"通过人工智能或同等技术生成或操纵的合成内容"，其逼真程度足以"改变任何活人、已故或虚构人物的形象、声音或言论"。1
这一定义是在处理一起涉及参议员弗拉维奥·博尔索纳罗的案件后确立的。1参议员博尔索纳罗在7月自由党党代大会上使用了人工智能生成的被监禁父亲、前总统博尔索纳罗的形象来支持其竞选活动。1前总统博尔索纳罗于11月因领导政变企图被判处27年监禁。1
法院最终以4:3的投票结果裁定不对参议员进行处罚，认定该深度伪造不构成公开政治广告，而是属于私人事件性质。1然而，法院随后以5:2的投票通过了关于政治广告中深度伪造的标准。1这些规则将适用于10月4日首轮投票前的大量竞选内容。1根据现行规定，广播电视必须每天预留14分钟播放约1000个30秒的广告，另外还有12.5分钟的时间段专门保留给总统候选人。1
Brazil's highest electoral court issued new guidelines on Tuesday to regulate artificial intelligence-generated deepfakes in political advertisements.1 The new standards define deepfakes as "synthetic content generated or manipulated through artificial intelligence or equivalent technology" with a level of realism sufficient to "alter the image, voice, or statements of any living person, deceased person, or fictional character."1
The ruling emerged from a case involving Senator Flavio Bolsonaro, who used an AI-generated image of his imprisoned father, former President Jair Bolsonaro, at a July party convention to support his campaign activities.1 The former president was sentenced to 27 years in prison in November after being convicted of leading an attempted coup.1 The court voted 4-3 to decline imposing penalties on the senator, determining that the deepfake did not constitute a public political advertisement but rather a private event.1 However, in a separate 5-2 vote, the court defined standards for deepfakes that appear in political advertising, which will apply to substantial campaign content circulating before the October 4 first round of Brazil's presidential election.1 Under current regulations, broadcast television must reserve 14 minutes daily for approximately 1,000 thirty-second advertisements, with an additional 12.5 minutes designated specifically for presidential candidates.1
评论
还没有评论，欢迎留下第一条。