一项由美国国会监督机构开展的调查发现,包括聊天机器人在内的人工智能模型在测试中被用于协助策划针对美国的恐怖袭击计划。1 尽管这些平台设有安全防护措施,但政府调查人员在卧底行动中仍发现它们能够生成“可行”的袭击方案。12 这一发现凸显了当前人工智能监管存在的不足,并直接促使美国参议院着手筹备相关听证会。12
针对这一安全隐患,各方呼吁立即采取行动以应对潜在风险。2 《纽约时报》、《华盛顿邮报》及福克斯新闻等多家媒体均报道了政府调查人员发现AI平台在被要求时产出恐怖阴谋计划的事实。2
A congressional watchdog agency conducted an undercover investigation revealing that American artificial intelligence models, including chatbots, have the potential to be turned against the United States by assisting in the planning of terrorist attacks. 1 Despite existing safeguards on these platforms, government investigators found that they could generate "viable" attack plans when prompted during tests. 12 This discovery has raised concerns about inadequate current regulation of AI. 1
In response to these findings, the U.S. Senate is preparing to hold hearings focused on artificial intelligence oversight. 12 The urgency of the situation has prompted calls for immediate action, with officials stating, "We need to act now." 2 Multiple media outlets, including CBS News, reported that the investigation demonstrates how readily these technologies could be exploited if safeguards are not strengthened. 2
评论
还没有评论,欢迎留下第一条。