Reporting from Zenith Live 2025: : Advancing Data Security with AI and LLMs

This is part of a series reporting live from Zenith Live 2025, where Zscaler is unveiling groundbreaking innovations that reshape secure healthcare IT. Today, I’m focusing on two critical advancements in data security designed to help healthcare providers harness the power of AI and large language models (LLMs) while keeping sensitive data protected and workflows compliant.

Securing Private AI and LLMs with Zscaler AI Guard

As healthcare organizations increasingly explore private AI models—whether for clinical decision support, patient-facing chatbots, or operational efficiencies—the risks of unsecured or poorly controlled deployments have taken center stage.

Enter Zscaler AI Guard, a new capability built to secure private AI and LLM deployments. This feature provides comprehensive security controls for AI models, ensuring sensitive data remains protected, inappropriate outcomes are mitigated, and workflows are continuously monitored. With AI Guard in place, you can be confident that your AI model won’t be “jailbroken,” exposing sensitive healthcare data or generating unintended results.

For healthcare providers, this means confidently adopting private AI solutions with guardrails that mitigate reputational risks, regulatory violations, and patient data exposure. Whether piloting a medical documentation assistant or developing advanced patient engagement tools, Zscaler AI Guard ensures these innovations are secure, compliant, and trustworthy from day one.

But AI security is just one facet of a larger challenge. The data healthcare organizations rely on spans cloud platforms, on-premises systems, and legacy databases—making broad, integrated security essential.

Extending Data Security with DSPM and On-Prem Database Scanning

Zscaler’s Data Security Posture Management (DSPM) has been expanded to include agentless scanning of on-premises databases, covering Oracle, MSSQL, MySQL, and PostgreSQL. This approach enables swift data classification while minimizing performance impact through sample-based scans run directly from the customer’s cloud environment. With DSPM, healthcare organizations gain clear visibility into the security posture of their on-premises data, identifying exposure risks without requiring intrusive agents or workflow changes.

For healthcare providers, this capability protects sensitive operational data—like patient care records, billing information, and internal workflows housed in EHR systems. DSPM’s expanded reach ensures comprehensive protection for data environments across both cloud and on-premises systems.

AI-Classified Data Protection and AI SPM

Traditional data classification methods relying on static rules and regex are no longer adequate to protect sensitive healthcare data. Zscaler’s new AI-Classified Data Protection takes a dynamic approach, identifying and classifying sensitive data in real time by understanding the content and context of each data element. This enables healthcare IT teams to move beyond rigid rulesets toward adaptive data protection that evolves with changing regulatory landscapes.

Additionally, Zscaler is introducing AI Security Posture Management (AI SPM), a solution focused on securing AI workloads and preventing AI-driven data leaks. AI SPM assesses AI resources—including models, tools, and agents—against risks such as misconfigurations, supply chain vulnerabilities, and exposure of enterprise data to AI systems. It correlates AI-related data risks to deliver rapid insights into compliance gaps and exposure risks.

Why This Matters for Healthcare IT

The AI and DSPM capabilities unveiled at Zenith Live 2025 empower healthcare IT to become proactive partners in innovation—enabling technological progress without compromising security.

• Secure private AI models with guardrails that prevent data leaks and inappropriate outcomes.
• Strengthen real-time data protection through adaptive AI-driven classification tailored to complex healthcare data and compliance requirements.
• Extend DSPM visibility to on-premises databases, ensuring critical operational data is classified and protected without disrupting workflows.
• Identify and govern AI-related risks while adopting AI compliance frameworks through AI SPM.

Stay tuned for more updates from Zenith Live 2025, where I’ll continue to cover the latest innovations shaping the future of secure healthcare IT.