伊朗一所学校的炸弹袭击事件可能与人工智能错误有关。
AI Error May Have Contributed to Girl's School Bombing in Iran

原始链接: https://thisweekinworcester.com/exclusive-ai-error-girls-school-bombing/

伊朗联合国大使称,伊朗米纳布一所女校最近遭受导弹袭击,据报道造成150名学生死亡,但这一数字尚未得到确认。调查显示,美国军方可能对此负责,并非故意为之,而是由于其快速扩展的人工智能使用中的错误。 司法部和国防部内部消息人士表示,该人工智能系统,一个基于Claude的程序,可能使用了过时的情报,将学校定位在伊朗伊斯兰革命卫队(IRGC)前据点附近。五角大楼正在调查发射是如何被授权的。 此事件紧随之前有关人工智能错误导致埃普斯坦文件发布问题的报告。值得注意的是,特朗普政府最近将Claude的创建者Anthropic列为供应链风险,原因是担心政府监控,导致转向OpenAI的ChatGPT。军方有六个月的时间逐步淘汰Claude。

一份来自*thisweekinworcester.com*的报告称,人工智能程序(很可能是Anthropic的Claude)与伊朗最近炸死女学生的事件之间可能存在关联。该理论认为,人工智能提供了过时的情报,包括学校的位置作为目标。 Hacker News上的讨论对此表示高度怀疑,许多评论者指出缺乏具体证据,并指责相关方猜测或试图转移责任。一些人认为该事件类似于特斯拉Autopilot的问题——尽管有警告,但过度依赖人工智能。另一些人认为,人工智能错误*和*人为意图都可能存在,使得问责变得困难。 一些评论集中在潜在动机和责任上,提及美国军方和政治人物过去的行动。人们也对公司将人工智能作为替罪羊以及事件缺乏透明度表示担忧(“我们不知道黑匣子里有什么”)。最后,一些用户质疑原始新闻来源的可信度。
相关文章

原文

This Week in Worcester spoke with multiple sources who confirmed that the military’s deployment of AI led to the missile strike against the Shajareh Tayyebeh girls’ school in Minab, a city in southern Iran, on Saturday. Ali Bahreini, Iran’s ambassador to the U.N. in Geneva, said ​the strike killed 150 students. There has been no independent confirmation of that death toll.

The Pentagon is currently investigating the matter, and according to a report from Reuters, officials within the military confirmed the US’s potential responsibility for the attack. There is no evidence at this time that the U.S. military intentionally targeted the school. A compound near the school was previously associated with the Islamic Revolutionary Guard Corps (IRGC).

This Week in Worcester spoke with a Department of Justice appointee on the condition of anonymity, citing an ongoing and active investigation. “The immediate theory is that the AI program included the school’s position based on older, archived intelligence. The logic behind the launch, and the mechanics of who authorized it is unclear.”

This Week in Worcester spoke with a logistics programmer in the Department of Defense (DOD), who said that the department rapidly scaled up its use of a Claude-based system over the past year, integrating it with many core operational decisions.

“They are gung-ho about this program, and want to use it for everything. Most of their operational planning is done using this software, although there is some things we have designed in-house,” said the appointee.

The incident in Iran is currently under investigation by military investigators.

Press Secretary Karoline Leavitt said in a statement to Reuters, “While the Department of War is currently ​investigating this matter, the Iranian regime targets civilians and children, not the United States of America.”

Defense Secretary Pete Hegseth said in response to being asked about the incident: “We’re investigating that. ​We, of course, never target civilian targets. But we’re taking a look and investigating that.”

Multiple news agencies reported on the military’s use of Claude AI, made by Anthropic.

This week, the Trump Administration declared Anthropic a supply chain risk over the company demanding that the government not use its technology for mass surveillance of Americans or autonomous vehicles. The military has six months to eliminate Claude usage. The administration signed a contract with OpenAI, the maker of ChatGPT, after the announcement about Claude.

This Week in Worcester previously reported that AI errors led to delays in the release of the Epstein files, with many files being either mistakenly redacted or unredacted without human supervision. The same source there says that some of the files being moved on and off the DOJ’s Epstein website are, “being double-checked by human attorneys.”

 

Image Credit: Touch Of Light, The Pentagon, Headquarters of the US Department of Defense (cropped2), CC BY-SA 4.0

联系我们 contact @ memedata.com