The National Academy of Medicine has concluded that decades-old patient safety frameworks are inadequate for an AI-driven healthcare system, and the organization is now drafting new conditions to ensure artificial intelligence makes care safer rather than riskier. The effort, published on July 13, 2026, represents the first comprehensive attempt to align patient safety strategy with the reality of AI adoption in hospitals and clinics.

What Changed in Healthcare Safety Since the Last Major Framework?

For 26 years, patient safety efforts in the United States have been guided by principles established in the landmark 2000 report "To Err Is Human." But the healthcare landscape has transformed dramatically. Treatment options have multiplied, care systems have become far more complex, and the medical knowledge base has expanded exponentially. Digital infrastructure and artificial intelligence are now reshaping what patients and clinicians can do, expect, and are responsible for.

The National Academy of Medicine reviewed patient safety frameworks from the Organisation for Economic Co-operation and Development (OECD), the World Health Organization (WHO), and individual countries including the United Kingdom, Australia, Canada, the Netherlands, Ireland, and Norway. All of these frameworks emphasized leadership and culture, policy and governance, data transparency, patient engagement, teamwork, and capacity building. But they shared a critical blind spot: none were developed with generative AI in mind, and none treated AI as a mechanism to advance patient safety at scale.

How Can AI Actually Make Healthcare Safer?

Rather than viewing AI purely as a risk to manage, the National Academy of Medicine's initiative frames artificial intelligence as a tool to dismantle persistent barriers to patient safety. AI offers capabilities that were previously impossible: integrating diverse clinical, behavioral, environmental, and experiential information in real time; accurately measuring patient outcomes; proactively identifying which patients face the highest risk of harm; and enabling real-time intervention when problems emerge.

The organization notes that traditional patient safety improvements have yielded uneven results. Gains have been concentrated in pockets of excellence rather than spread system-wide, and they have been difficult to sustain and scale. The overall rate of patient harm has remained stubbornly resistant to change. AI, combined with a deliberate national strategy, could unlock cascading improvements across the entire health ecosystem.

Steps to Integrate AI Safely Into Healthcare Systems

• Establish Essential System Conditions: The National Academy of Medicine is developing draft essential conditions that healthcare systems must meet before deploying AI, moving beyond simply managing AI safety to using AI to make care safer for all patients.
• Create Independent Oversight: All reviewed national strategies had an independent national patient-safety agency. The United States will need similar governance structures specifically designed for AI-enabled care.
• Prioritize High-Leverage Actions: Rather than attempting comprehensive transformation all at once, the strategy focuses on a limited set of high-impact actions that can unlock broader opportunities across hospitals, clinics, research institutions, and the wider health ecosystem.
• Build a Learning Health System: AI should enable continuous monitoring, adaptation, and learning across the entire healthcare network, advancing the shared commitment to a national learning health system that evolves based on real-world outcomes.

Why Existing Safety Frameworks Fall Short for AI

The frameworks reviewed by the National Academy of Medicine presented their core components in largely linear models that did not reflect the dynamic interactions among them. Each country's approach was also shaped by its unique health system structure, limiting direct applicability to other contexts. More fundamentally, all strategies were developed before the widespread adoption of generative AI and therefore did not address its implications for patient safety.

The organization concluded that while these frameworks serve as useful reference points, they cannot serve as a blueprint for a national strategy that accounts for AI. The United States needs a deliberately different approach that addresses gaps in prior frameworks while setting the stage for broader advances in system-wide precision and effectiveness of healthcare.

What Does the Laboratory AI Market Tell Us About Healthcare's AI Future?

Beyond clinical care, artificial intelligence is rapidly transforming the laboratory infrastructure that supports healthcare. The global laboratory AI market was valued at $2.18 billion in 2025 and is projected to reach $6.48 billion by 2033, growing at a compound annual rate of 14.60 percent. This explosive growth reflects the urgency with which pharmaceutical companies, clinical laboratories, and research organizations are adopting AI to accelerate drug discovery, improve diagnostic accuracy, and address persistent shortages of skilled laboratory personnel.

Machine learning (ML) led the laboratory AI market with a 34 percent share in 2025, deployed across laboratory automation, predictive analytics, molecular modeling, and experimental optimization. Generative AI (GenAI) is the fastest-growing technology type, projected to register a compound annual growth rate of 16.5 percent, reflecting the surge in use of large language models and generative foundation models for scientific research and laboratory automation.

The hardware equipment segment dominated the component category with a 35.6 percent revenue share in 2025, driven by increasing deployment of AI-integrated analytical instruments, imaging systems, robotic sample handlers, and automated laboratory platforms. Life sciences applications accounted for 54 percent of the market, with AI being adopted for drug discovery, genomics research, biomarker identification, diagnostic testing, and laboratory workflow automation.

North America dominated the laboratory AI market with the largest revenue share of 38.9 percent in 2025, supported by advanced pharmaceutical and biotechnology research infrastructure and widespread laboratory automation. However, Asia-Pacific is expected to be the fastest-growing region at a compound annual growth rate of 15.2 percent from 2026 to 2033, fueled by expanding pharmaceutical manufacturing and accelerating AI adoption across China and India.

What Are the Biggest Obstacles to Deploying Laboratory AI?

Despite rapid growth, significant barriers remain. A major restraint in the laboratory AI market is the complexity of validating AI-generated outputs and integrating AI platforms with existing laboratory information management systems, analytical instruments, and legacy data infrastructures. Laboratories operating in regulated environments face particular challenges in ensuring data integrity, traceability, and regulatory approval.

These technical and regulatory hurdles underscore why the National Academy of Medicine's effort to establish safety conditions for AI in healthcare is so timely. As AI becomes embedded in both clinical workflows and laboratory operations, the healthcare system needs clear, evidence-based guidelines for when and how to deploy these tools responsibly. The organization's draft conditions aim to ensure that AI integration advances patient safety rather than introducing new risks into an already complex system.