October 8, 2026 | Clearwater Healthcare Cyber Briefing

AI on Defense: What Healthcare Is Learning About AI Risk

Executive Summary

Healthcare AI adoption is moving faster than many organizations’ ability to discover, evaluate, and secure the technologies entering their environments. AI capabilities are expanding across clinical workflows, administrative operations, research, and cybersecurity, while governance programs struggle to maintain visibility into what these systems can access and do.

In Clearwater’s October Healthcare Cyber Briefing, Dave Bailey, VP of Consulting Solutions & Strategy, opens with the latest healthcare threat intelligence and regulatory developments. Harry Lu, VP of AI Solutions at Clearwater, then joins Mahesh Chintakunta, Deputy CISO at Rush University System for Health, to examine practical AI governance and the Coalition for Health AI (CHAI) cybersecurity playbooks they are helping develop.

The discussion also introduces findings from Clearwater’s 2026 Healthcare AI Security & Governance Benchmark Study, including a critical visibility gap: 91% of respondents reported being unable to validate or verify AI operating within their environments.

The central takeaway is that healthcare organizations need more than AI policies. They need the ability to discover AI, understand its access and decision-making authority, validate vendor controls, and demonstrate that governance works in practice.

In This Discussion

This briefing explores questions including:

• What AI-related cybersecurity threats should healthcare organizations be monitoring?
• What is the CHAI Defensive Playbook, and how will it help healthcare organizations?
• How is Rush University System for Health approaching AI governance?
• Why is shadow AI creating gaps in organizational visibility?
• What should healthcare organizations ask vendors about embedded AI?
• How can organizations maintain meaningful human oversight of AI agents?
• What does Clearwater’s 2026 Healthcare AI Benchmark reveal about governance maturity?

What Changed in Healthcare Cybersecurity Threats in October 2026?

Dave Bailey opens the briefing during Cybersecurity Awareness Month with a reminder that cyber safety is patient safety. As healthcare organizations adopt more AI-enabled technologies, maintaining clinical resilience requires understanding not only traditional cyber threats but also the dependencies and behaviors introduced by AI.

The Critical Threat Brief highlights three developments:

AI is becoming both a security tool and an attack surface. Dave discusses examples of autonomous AI agents attempting to bypass access restrictions. These incidents illustrate why organizations need technical safeguards that enforce boundaries, even when an AI system is working toward an otherwise legitimate objective.

Healthcare ransomware activity continues to grow. Clearwater’s monitoring identified 413 healthcare-related victims through the first week of October 2026, a 71% increase compared with the same point in 2025. Approximately 70 distinct threat groups had appeared in the monitoring data, demonstrating how quickly the ransomware ecosystem is changing.

HIPAA risk analysis remains central to enforcement. Dave reviews OCR’s September settlement with Ambry Genetics, bringing the 2026 total discussed in the briefing to nine enforcement actions and approximately $2.98 million in penalties. All nine actions involved missing or inadequate risk analysis.

The briefing reinforces that healthcare organizations should strengthen cybersecurity fundamentals while expanding governance to address emerging AI capabilities.

What Is the CHAI Defensive Playbook?

The Coalition for Health AI (CHAI) is developing offensive and defensive cybersecurity playbooks intended to help healthcare organizations address the risks associated with increasingly capable AI systems.

Harry Lu and Mahesh Chintakunta discuss their participation in this collaborative effort, which brings together healthcare providers, technology companies, consulting organizations, and other industry stakeholders.

The goal is to translate emerging AI security threats into practical guidance that healthcare organizations can use without having to develop an entirely new framework independently.

Mahesh describes the purpose of the Defensive Playbook:

Mahesh Chintakunta: “It’s going to bridge the gap between the excitement the organizations have around AI and the operational reality of securing it.”

The work draws on established cybersecurity and AI risk management approaches while addressing challenges specific to healthcare.

At the time of the October briefing, the working group was targeting publication of the offensive and defensive playbooks in December 2026, subject to review and approval.

How Is Rush University System for Health Governing AI?

Mahesh shares how Rush has established a formal, cross-functional AI governance structure.

The process includes representatives from the AI Center of Excellence, information security, privacy, compliance, legal, research, and operational teams.

When a proposed AI application involves patient data, research information, or other sensitive organizational information, it undergoes technical review and cybersecurity risk evaluation. Recommendations then move to the governance committee, where stakeholders evaluate the proposed use and associated risks.

Higher-risk implementations may require formal risk acceptance.

Mahesh emphasizes that cybersecurity should help the organization adopt AI responsibly, not prevent innovation.

Mahesh Chintakunta: “We are not a department of NO, we are a department of enabler.”

Three questions help guide Rush’s governance decisions:

1. What data is being used?
2. What decisions are being influenced?
3. What happens if the AI is wrong?

These questions connect technical security assessments with clinical, operational, and organizational accountability.

Why Is Shadow AI One of Healthcare’s Biggest Governance Challenges?

Even organizations with established governance committees may not know every AI application operating within their environments.

Mahesh identifies shadow AI and limited visibility into AI systems as two significant gaps between adoption and governance.

Employees can introduce AI through personal tools, approved platforms, embedded vendor features, or applications developed internally. Many of these capabilities never pass through traditional procurement or technology review processes.

Harry explains the value of establishing an AI front door, a structured process through which business teams can submit proposed AI use cases for evaluation.

However, a formal intake process alone is insufficient.

Organizations also need independent discovery capabilities to identify AI systems that bypass established governance channels.

The objective is not simply to count AI applications. Security teams must determine what those systems access, whether they process sensitive information, and how they interact with other enterprise technologies.

Harry highlights the importance of validating what organizations believe they know about their AI environments.

What Should Healthcare Organizations Ask AI Vendors?

As AI becomes embedded within software-as-a-service (SaaS) platforms, healthcare organizations may inherit new AI functionality without intentionally purchasing or implementing a separate AI product.

Some vendors allow customers to disable these capabilities. Others integrate AI directly into core platform functionality.

The discussion emphasizes that vendor oversight should extend beyond traditional functionality and data security questionnaires.

Healthcare organizations should understand what decisions AI can influence, what permissions it possesses, and whether its actions can be independently verified.

Mahesh raises a particularly important question:

Mahesh Chintakunta: “How much of decision-making are we willing to delegate to AI, and how do we govern that?”

This becomes increasingly important when AI is integrated into clinical operations, billing, administrative processes, and systems containing protected health information.

How Can Healthcare Organizations Maintain Human Oversight of AI Agents?

Human oversight remains a central theme throughout the conversation.

Mahesh explains that organizations may initially review AI-generated outputs carefully, but repeated satisfactory results can create excessive confidence in automated decisions.

Over time, human reviewers may become less attentive, even when formal policies still require human involvement.

This creates a risk that human oversight exists in theory but becomes ineffective in practice.

Healthcare organizations should define which decisions AI can make independently, which require human approval, and how AI-generated actions can be traced back to accountable individuals.

The discussion also highlights the importance of non-human identity management. AI agents may need access to sensitive information, enterprise applications, or privileged functions.

Organizations should be able to determine what an agent accessed, what actions it performed, and who authorized those activities.

What Does Clearwater’s 2026 Healthcare AI Benchmark Reveal?

Harry closes the featured discussion by highlighting three findings from Clearwater’s 2026 Healthcare AI Security & Governance Benchmark Study.

1. AI visibility and validation remain major challenges.

91%

of survey respondents reported that they could not validate or verify the AI operating within their environments.

This finding demonstrates the gap between knowing that AI is being adopted and independently confirming where it operates and what it can do.

2. Vendor assurance needs to improve.

Healthcare organizations are inheriting AI capabilities through third-party technologies. Security leaders need ways to validate vendor controls without creating unnecessary operational burdens.

3. AI security investment is not keeping pace.

Despite accelerating adoption and growing awareness of AI-related risk, dedicated security investment remains insufficient in many organizations.

The findings reinforce the need for better discovery, measurable security controls, and governance programs that can scale alongside AI adoption.

Key Takeaways

• AI is creating new operational and cybersecurity risks while also increasing the capabilities available to defenders and attackers.
• Healthcare ransomware activity remains elevated, and threat actors continue changing names, tactics, and targets.
• HIPAA enforcement continues to reinforce the importance of accurate, thorough risk analysis.
• The CHAI Defensive Playbook aims to translate emerging AI threats into practical cybersecurity guidance.
• Cross-functional AI governance should evaluate data access, decision-making authority, and the consequences of incorrect AI outputs.
• Shadow AI and embedded vendor capabilities make independent discovery and validation essential.
• Human oversight must remain meaningful as AI systems become more autonomous.
• Clearwater’s benchmark findings reveal significant gaps in AI visibility, vendor assurance, and security investment.

Questions This Discussion Answers

• What are the most significant AI cybersecurity risks facing healthcare organizations in 2026?
• What is the CHAI Defensive Playbook?
• How is Rush University System for Health approaching AI governance?
• What is an AI front door, and how does it support governance?
• Why is shadow AI difficult for healthcare organizations to manage?
• How should healthcare organizations evaluate AI embedded in vendor software?
• What does meaningful human oversight of AI agents require?
• How should non-human identities and AI permissions be governed?
• What does Clearwater’s 2026 Healthcare AI Security & Governance Benchmark reveal?
• How can healthcare organizations strengthen AI security without slowing innovation?

About This Healthcare Cyber Briefing

Clearwater’s Healthcare Cyber Briefing is a monthly webinar series examining developments in healthcare cybersecurity, privacy, compliance, and enterprise risk management.

In this October session, Dave Bailey delivers Clearwater’s Critical Threat Brief before Harry Lu and Mahesh Chintakunta explore the work underway through the Coalition for Health AI, practical approaches to healthcare AI governance, and the challenges of securing increasingly autonomous technologies.

The session also introduces findings from Clearwater’s 2026 Healthcare AI Security & Governance Benchmark Study, offering healthcare executives, CISOs, compliance leaders, and technology professionals practical perspectives on closing the gap between AI adoption and security readiness.

Watch the full October Healthcare Cyber Briefing replay to hear the discussion and explore how healthcare organizations can develop a more defensible approach to AI risk.