Two distinct Israeli initiatives, both relying on artificial intelligence, surfaced this week, revealing a coordinated push to apply machine-learning techniques to medicine and to statecraft. The neurotechnology startup Hemispheric emerged from six years of stealth with $52 million in funding and a foundation model trained on more than 250,000 hours of electroencephalogram recordings from over 100,000 participants across laboratories in Israel, the United States and the Philippines. Separately, documents and reporting confirmed that Israel’s government has committed at least $45–50 million to a digital influence operation targeting US audiences, using AI-generated text messages and content networks to counter what officials in Jerusalem describe as a sharp decline in American public support.

Hemispheric’s model, named Descartes, processes the complex electrical language of the brain through a deep-learning architecture that accounted for six billion parameters during training. The company’s chief technology officer, Gidi Littwin—a co-inventor of the core component behind Apple’s Face ID—and CEO Hagai Lalazar, a brain researcher, argue that existing EEG technology, though a century old, has remained clinically impractical because its waveforms vary so greatly between individuals that human interpretation is unreliable. The model is currently being tested on patients with post-traumatic stress disorder, schizophrenia and depression, and the company plans to submit its first product, a PTSD diagnostic aid, to the US Food and Drug Administration early next year, with a target market entry by the end of 2027. Viewed from East Asian technology circles, the venture is being watched as a test case for whether the data-engineering discipline honed in consumer hardware can be ported into regulated medical devices.

The influence campaign, by contrast, is already live. Operated by a firm owned by former Trump strategist Brad Parscale, it has flooded American mobile phones with messages from fictitious personas such as “Emma” or “Sarah” in a group called Friends for Peace, asking recipients about the security implications of US-Israel-Iran diplomacy. Analysts in Jakarta note that the operation also builds pro-Israel websites designed to be ingested by large language models like ChatGPT, so that when users query AI chatbots about the bilateral alliance, the answers draw on curated sources. The Israeli foreign ministry has set a 2026 public-diplomacy budget exceeding $700 million, a figure that researchers at the Quincy Institute in Washington say would place Israel among the highest-spending nations on disclosed influence efforts in the United States this year.

Both projects share a structural challenge: data scarcity in their respective domains. Hemispheric had to build its own proprietary brain-activity database from scratch because no suitable corpus existed for training a foundation model on neural signals. The influence campaign, meanwhile, confronts a different kind of scarcity—dwindling political sympathy—and attempts to compensate by generating synthetic conversational scale through AI. In Tehran, state-affiliated media have framed the influence operation as evidence of a broader effort to manipulate American public opinion, while Israeli business outlets present Hemispheric’s technology as a potential exportable diagnostic platform. The startup is already selling its know-how to pharmaceutical companies for use in clinical research, and its founders aim to shrink the hardware cost enough to place EEG-based screening in general practitioners’ offices.

The next concrete milestones are regulatory and legislative. Hemispheric’s FDA submission for its PTSD product is expected in early 2027, a process that will test whether an AI model trained on a global dataset can satisfy US medical-device standards. On the influence front, the 2026 Israeli state budget’s allocation for public diplomacy will be scrutinised as the campaign’s scale becomes more visible to American regulators and platforms.