以色列AI双线推进：脑电解码进入临床前夜，同时被曝千万美元影响美国舆论
一家由苹果Face ID发明人联合创办的初创公司完成5200万美元融资，利用10万人脑电数据训练诊断模型；同期，以色列政府被指斥资逾4500万美元，通过AI生成信息试图扭转美国民意。
以色列在人工智能应用领域同时展现出两条截然不同的轨迹。脑科学初创公司Hemispheric在隐匿运营六年后，于本周宣布累计融资5200万美元，并公布其核心模型“笛卡尔”（Descartes）的训练细节。该模型基于超过10万名受试者、总计25万小时的脑电图数据，旨在将大脑电活动转化为可供临床医生解读的信息，以辅助诊断抑郁症、创伤后应激障碍（PTSD）、帕金森病及早期认知衰退。公司联合创始人兼首席技术官Gidi Littwin曾是苹果Face ID核心专利的共同持有人，他将此前处理大规模数据的经验带入了这一项目。
Hemispheric自2019年起在以色列、美国和菲律宾设立实验室，让志愿者佩戴轻便头戴设备进行15分钟的脑电记录。模型训练时考虑了60亿个参数，目前已在PTSD、精神分裂症和抑郁症患者中展开测试，并显示出较高的判读准确率。公司计划明年初向美国食品药品监督管理局（FDA）提交首款针对PTSD的产品申请，目标在2027年底前推向市场。首席执行官Hagai Lalazar表示，长远愿景是让脑健康检测像常规血液检查一样普及，为此将压低设备成本，使其能进入心理诊所和普通医院。
与此同时，以色列政府被曝投入巨资试图影响美国公众舆论。《华尔街日报》报道称，以色列内阁与曾任特朗普战略顾问的Brad Parscale旗下公司签署了价值超过4500万美元的合同，通过AI生成数百万条短信和构建亲以色列内容网站，以改善其在美形象。皮尤研究中心3月的民调显示，约60%的美国成年人对以色列持负面看法，加沙战争和与伊朗的紧张关系是主要因素。部分美国民众已察觉到这些AI对话的机械感，而前特朗普策略师Steve Bannon则评论称，这一努力是在“试图推销无法推销的东西”。
两条线索共同勾勒出以色列在AI领域的深度介入：一边是医疗科技向临床转化迈进，另一边则是数字影响力操作引发争议。Hemispheric的下一步关键节点是FDA的审批进程，而以色列外交部已为2026年编列逾7亿美元的全球形象预算，预示着此类行动可能进一步扩大。
| 伊朗及相关媒体 | −0.80 | critical |
|---|---|---|
| 以色列媒体 | −0.20 | neutral |
| 大西洋／英语圈媒体 | +0.50 | aligned |
| 东南亚媒体 | 0.00 | neutral |
The Zionist regime orchestrates a media manipulation campaign with AI to buy American consent.
By using the term 'plot' and emphasizing the secretive nature and the large sum of money, the narrative creates a sense of threat and moral outrage, positioning Israel as a deceptive actor.
The Iranian frame omits the existence of the Israeli startup Hemispheric and its legitimate medical AI research, which is part of the same story, thereby presenting a one-sided view of Israel's AI activities.
Israel admits the failure of its public relations campaigns, but at the same time claims the success of its technological innovation in neuroscience.
By juxtaposing a critical report on government spending with a celebratory startup story, the Israeli press creates a balanced but ultimately self-justifying narrative: the state's efforts may be ineffective, but Israeli ingenuity still shines.
The Israeli frame omits the connection between the two stories: the fact that the same AI technology used for medical research is also being deployed for influence operations, which could raise ethical questions.
The Israeli startup Hemispheric revolutionizes neurological diagnostics with AI that deciphers brain activity, promising diagnoses as simple as a blood test.
By emphasizing the credentials of the founders (Apple Face ID inventor) and the large funding, the narrative builds credibility and frames the technology as a legitimate scientific advancement, avoiding any political context.
The Atlantic frame omits the other half of the story: the Israeli government's AI-driven influence campaign in the US, which is also part of the headline. This omission allows the narrative to remain purely scientific and apolitical.
Israel spends 731 billion rupiah on a digital influence campaign in the United States, using AI to send messages to millions of Americans.
By using precise numbers and describing the operation in a straightforward manner, the narrative achieves an appearance of objectivity, while implicitly normalizing the use of AI for political influence.
The Southeast Asian frame omits the medical AI startup story, thus presenting the Israeli AI activities solely as a tool for propaganda, without the counterbalance of scientific innovation.