In both cases, Israel is offering an asset that is difficult to replicate elsewhere: a centralized, highly digitized healthcare system containing decades of medical information that allows new technologies to be tested and implemented at scale.
During the pandemic, Israel became a real-time testing ground for evaluating vaccine effectiveness. Today, it is positioning itself as a testing ground for the next generation of medical artificial intelligence.
1 View gallery
Sam Altman alongside Sheba.
(Photos: Anna Moneymaker/Getty Images, Sheba)
The comparison, however, has limits. The Pfizer agreement was aimed at addressing a global health crisis through vaccine deployment and research, while the OpenAI partnership focuses on developing and implementing AI tools that could become part of the daily workflow of doctors and medical teams.
Yet both agreements raise similar questions: What value does an international company gain from access to Israel’s healthcare system? What benefits do patients and the healthcare system receive? And where should the line be drawn between medical innovation and concerns over privacy, data security, and dependence on commercial technology providers?
At the center of the collaboration are Israeli patients.
The asset attracting AI companies and pharmaceutical giants to Israel is not only the country’s medical expertise or startup ecosystem, but also its digital healthcare infrastructure. Over decades, Israel’s healthcare organizations have accumulated extensive medical records, creating one of the most comprehensive and consistent healthcare datasets available for research and technology development.
For companies developing medical AI systems, such data is essential for training, calibrating, and validating models.
As part of the agreement, Sheba will become the first hospital outside the United States to receive access to OpenAI’s dedicated healthcare AI models, including ChatGPT Health.
For OpenAI, the partnership provides an opportunity to integrate institutional knowledge, including clinical protocols, treatment pathways, and hospital procedures, into AI systems designed for healthcare environments.
For Sheba, which has operated the ARC Innovation Center since 2019, access to these tools is intended to support doctors and medical teams by assisting with clinical decision-making, comparing recommendations with medical research, and reducing the administrative burden involved in documentation.
The potential impact of AI in healthcare extends beyond answering medical questions.
Hospitals are increasingly looking at AI as a tool to automate time-consuming administrative processes, including preparing medical documentation, analyzing clinical information, processing authorization requests, and interpreting complex datasets.
If successfully implemented, these systems could shorten waiting times, improve continuity of care, and support more personalized medicine.
They could also allow patients outside major medical centers to benefit from recommendations based on broader global medical knowledge.
As part of the Sheba agreement, AI models are expected to be integrated into hospital protocols and potentially provide recommendations tailored to the characteristics of Israeli patients and the local healthcare environment.
The agreement comes as technology companies compete to develop AI systems specifically designed for medicine.
Anthropic, for example, has launched healthcare and life sciences packages focused on integrating AI into medical organizations. Unlike systems designed primarily as patient assistants or physician advisors, these tools are intended to interact directly with healthcare infrastructure, including information systems and administrative workflows.
Potential applications include automating insurance authorization requests, processing laboratory data, and drafting medical records for review by physicians.
The same medical information that makes Israel attractive to AI companies also creates its biggest challenge.
The principle established over decades is that medical information belongs primarily to the patient, not to doctors, hospitals, healthcare providers, or the government.
Even when information is anonymized and transferred for research purposes, questions remain over whether commercial companies should be allowed to use it without explicit patient consent and whether such practices could weaken public trust in the healthcare system.
Those questions were raised during Israel’s vaccine agreement with Pfizer, when critics questioned aspects of medical data sharing and the confidentiality of some agreement details.
The debate is now returning with greater intensity because AI systems are not only being used for research. They could become part of the infrastructure on which future healthcare depends.
Concerns include whether sensitive medical information, including test results, medical records, and imaging data, could be used to train commercial systems without sufficient transparency, or whether such information could become vulnerable in future cyberattacks.
The privacy debate is accompanied by growing concerns over cybersecurity.
The Ministry of Health warned last March of rising cyber risks linked to the use of public AI tools and identified them as a significant potential attack vector. As a result, the ministry blocked access to external AI tools from government hospital computers.
However, blocking access does not necessarily prevent usage. Many healthcare professionals continue to use external AI tools through personal devices during their work, sometimes sharing screenshots, medical documents, or imaging results to receive quick assistance.
This creates a challenge: without secure, approved systems, sensitive medical information may move through less controlled channels.
The Sheba-OpenAI agreement represents a major step in Israel’s effort to integrate artificial intelligence into healthcare. But the success of the partnership will depend not only on technological capabilities, but also on whether hospitals, regulators, and patients can establish trust around how medical information is used.
Israel’s digital healthcare system may become one of the world’s most valuable assets in the AI era. The next challenge will be determining how to use that asset while protecting the people whose data created it.