Israel's Largest Health System Builds AI Security Gateway to Stop Patient Data Leaks Through ChatGPT

Israel's largest health maintenance organization is racing to solve a problem facing healthcare systems worldwide: how to let clinicians use powerful AI tools like ChatGPT without accidentally leaking sensitive patient information into the public domain.

Clalit Health Services, which serves 5 million patients across Israel, has entered a multimillion-shekel strategic partnership with Check Point Software Technologies to develop an AI security gateway designed to prevent healthcare staff from inadvertently entering patient data into public generative AI platforms. The system, expected to be fully deployed by the first quarter of 2027, represents one of the first large-scale attempts to balance the productivity promise of generative AI with stringent healthcare privacy requirements.

Addressing an Escalating Threat Landscape

The partnership comes as Clalit faces an intensifying cybersecurity environment. According to Adv. Avivit Kotler, Clalit's Chief Information Security Officer, the organization experiences tens of thousands of cyberattacks annually, with rates "hundreds of times higher than usual" since the outbreak of war, particularly from Iran and its proxies.

The initiative also follows a ransomware attack on Hillel Yaffe Medical Center and subsequent tightening of cybersecurity regulations by Israel's Ministry of Health. The primary risk scenario is straightforward but consequential: healthcare workers seeking to leverage AI productivity tools might copy patient information into platforms like ChatGPT or Gemini, inadvertently exposing protected health data to public systems.

Beyond preventing data leakage, the security gateway will also defend against attacks from autonomous AI agents, addressing emerging threats as artificial intelligence systems become more sophisticated and potentially adversarial.

Balancing Innovation and Protection

Clalit already operates AI systems in clinical settings, including the Aidoc platform integrated into internal imaging systems. According to a World Health Organization report, the organization also uses local models to scan historical data for previously undiagnosed diseases. The new security infrastructure aims to extend these capabilities by enabling safe use of public generative AI tools rather than imposing blanket bans.

Under the partnership terms, Check Point will access only metadata rather than patient data itself, maintaining privacy protections while developing the security framework. Check Point retains intellectual property rights for future commercialization, positioning the cybersecurity company to market healthcare-specific AI security solutions to other health systems confronting similar challenges.

A Blueprint for Healthcare AI Adoption

The Clalit-Check Point collaboration establishes a potential model for healthcare organizations worldwide struggling with the same tension between clinician demand for AI productivity tools and institutional data security requirements. Rather than prohibiting generative AI use entirely, the security gateway approach enables controlled adoption that could improve clinical efficiency while maintaining compliance with privacy regulations.

For Check Point, developing and testing AI security infrastructure in a live environment serving 5 million patients represents a significant strategic opportunity. As healthcare systems globally move from isolated AI deployments to enterprise-wide generative AI adoption, demand for specialized AI security solutions is likely to accelerate.

The success or failure of this deployment will be closely watched by health systems, regulators, and technology companies as a test case for whether sophisticated security frameworks can reconcile the competing imperatives of AI innovation and patient privacy protection at scale.

Sources

1. Clalit and Check Point team up to secure AI use across Israel's largest HMO — Calcalist