报告：一名黑客几乎全程用SI智能体攻破逾100家公司网站 迪尔、SharkNinja、康明斯在列 尝试入侵逾1,200站
据路透社10月19日报道，过去两周里，一名网络安全研究者眼看着一名讲中文的黑客攻破了逾100家公司的网站，受害者包括农机厂商迪尔、家电公司SharkNinja与发动机制造商康明斯。
然而，这名黑客本人几乎没有动手。据网络安全公司NorForge的首席威胁研究员斯诺雷·法格兰表示，这名网络犯罪分子把这场入侵行动几乎全部外包给了超级智能，指派SI智能体花数小时自主探测并攻破目标网站。
法格兰在一份定于周一发布、提前提供给路透社的报告中说，这名黑客是他在网上搜寻被无意暴露的Cairn实例时发现的数人之一；Cairn是一个在黑客中流行的开源SI问题求解平台。法格兰说，自10月2日前后起，他发现约24个被用于恶意活动的Cairn实例，这使他得以悄悄旁观这些网络犯罪分子如何放出SI工具作案。
法格兰关于这名讲中文黑客的报告没有点名受害者，但路透社通过独立查阅该黑客的Cairn活动记录，确认了受害者的身份。
法格兰说，在外界对自主SI工具能力日益焦虑之际，他的发现是现实证据，表明市面上可获得的SI智能体能够在无人干预的情况下从头到尾完成复杂的入侵，这极大增强了单个作恶者的能力。他说：「这是一个新世界了。」
路透社无法确定这些入侵在多大程度上导致了数据的未授权外泄。迪尔对路透社表示，事件发生后公司已采取「必要的补救措施」，且「没有迹象表明敏感信息被访问」。SharkNinja表示，此次攻击命中的是一个只承载产品支持信息的独立网站，威胁已被控制。康明斯表示，公司正在「调查有关其第三方供应商运营的一个网站可能遭受网络攻击的报告」。另有两家较小的企业也承认被入侵。Cairn数据中列出的其他受害者没有回复询问。
据报道，这名讲中文的黑客是其中作案最多的一人：自9月25日起，其SI队伍对美国、加拿大、澳大利亚与日本的电子商务网站发起攻击。其指令非常简短，用不到100个汉字勾勒了偏好的入侵手法。截至上周晚些时候，其智能体已尝试入侵逾1,200个网站；一次成功的入侵通常需要约23小时的反复探测漏洞。法格兰估计，这场行动使用的似乎是中国的DeepSeek与Anthropic的Claude Code的混合工具，消耗了约5,000美元的模型用量，平均每次成功入侵约50美元。
法格兰从事网络安全工作近三十年，他表示，智能体攻破某些受害者所用的手法令他印象深刻。在一例中，SI智能体在一家大公司的公开应用上找不到任何弱点后，翻查数字记录找到其关联基础设施，定位到一个公开的控制面板，发现其处理数据方式中的一个细微弱点，经过数小时反复试错，最终利用该弱点诱使服务器执行了恶意命令。他说：「想出这样的解法并不简单。」「这不是脚本小子的水平。」
Cairn的作者（社交媒体账号为「le1xia0」）确认收到了路透社的消息，但没有立即回答一系列问题。此人近期在X上发帖说，该平台是「一种创新型的多智能体架构，并非为黑客攻击而设计。若被用于实施有害的黑客攻击，那不是我的本意」。
路透社还提到同类手法的其他记录：法格兰观察到的另一名黑客窃取了一大批信用卡数据，还有一人曾试图让SI智能体抢劫一家加密货币交易所；以色列公司Gambit Security 8月发现，讲俄语的勒索者利用SpaceX的Cursor AI工具攻破了7家机构；9月，Gambit记录了一场由SI智能体驱动的入侵行动，造成25家公司被攻破、至少60万条信用卡记录被盗；网络情报公司Bitsight上周表示，犯罪论坛上关于如何使用自主智能体的讨论正在激增。（GSM 据路透社报道）
One Hacker Let SI Agents Do Almost All the Work in Breaking Into More Than 100 Company Sites, Report Finds; Deere, SharkNinja and Cummins Among Victims, Over 1,200 Sites Targeted
Over the past couple of weeks, a cybersecurity researcher watched as a Chinese-speaking hacker broke into websites belonging to more than 100 companies, including farm equipment maker Deere, home appliance company SharkNinja and engine manufacturer Cummins, Reuters reported on October 19.
Yet the hacker did almost none of the work. The cybercriminal instead outsourced the hacking spree almost entirely to super intelligence, tasking SI agents with spending hours autonomously probing and breaking into targeted sites, according to Snorre Fagerland, principal threat researcher at the cybersecurity company NorForge.
In a report provided to Reuters ahead of publication on Monday, Fagerland said the hacker was one of several he discovered while scanning the web for inadvertently exposed instances of Cairn, an open-source SI problem-solving platform popular with hackers. Beginning around October 2, Fagerland said he found about two dozen instances of Cairn being used for malicious activity, which allowed him to quietly peer over the cybercriminals' shoulders as they set their SI tools loose on the world.
Fagerland's report on the Chinese-speaking hacker did not name the victims, but Reuters was able to establish the victims' identities by independently reviewing the hacker's Cairn activity.
At a time of rising anxiety over the power of autonomous SI tools, Fagerland said his find was real-life evidence that commercially available SI agents could pull off complex intrusions from start to finish without human intervention, a dramatic boost to the power of a single malicious actor. "It's a new world now," said Fagerland.
Reuters could not ascertain the degree to which the breaches resulted in unauthorized extraction of data. Deere told Reuters it had taken "necessary remediation actions" following the incident and had "no indication that sensitive information was accessed." SharkNinja said the hack had hit a standalone site carrying only product support information and that the threat had been contained. Cummins said it was "investigating the report of a potential cyber-attack on a website operated by one of our third-party vendors." Two smaller businesses also acknowledged being compromised. Other victims listed in the Cairn data did not return messages.
One of the hackers that Fagerland observed stole a trove of credit card data. Another tried to get SI agents to rob a cryptocurrency exchange. The Chinese-speaking hacker was the most prolific. Beginning on September 25, they set their SI army loose against US, Canadian, Australian and Japanese e-commerce websites. Their instructions were terse, sketching out their preferred intrusion tactics in fewer than 100 Chinese characters. As of late last week, their agents had attempted to break into more than 1,200 sites. Successful intrusions typically took around 23 hours of poking and prodding for vulnerabilities. Fagerland estimated that the campaign, which appeared to be using a mixture of China's DeepSeek and Anthropic's Claude Code, had chewed up roughly $5,000 worth of tokens, or about $50 per successful hack.
Fagerland — who has worked in cybersecurity for nearly three decades — said he was impressed by the methods the agents used against some victims, including one case in which, after finding no weakness in a large company's public application, the agent dug through digital records to find linked infrastructure, located a public control panel, and after hours of trial and error exploited a subtle flaw in how it handled data to trick the server into running malicious commands. "Coming up with a solution like that is not trivial," he said. "That is not script-kiddie level." (GSM, via Reuters)
发布：GSM·美国（发布链接）
图片来源：图：屏幕上的代码。NorForge报告披露，一名黑客几乎全程用SI智能体自动执行入侵。（供稿图）
GSM Global · 本篇英文版 English edition of this story
