And critically, neither governments nor industry have a plan to manage it. While the essay has far-reaching implications for all industries and humanity generally, three key risk categories emerge for health care and life sciences organizations. It is a useful moment to connect his framing to what is actually showing up in legislative bodies, enforcement actions, and litigation.
Labor Displacement Is No Longer a Future-Tense Problem
Gates first contends that AI-driven job displacement will move faster and cut deeper than prior technology transitions, in part because AI adapts to existing workflows rather than requiring the decades of infrastructure buildout required by prior revolutionary periods, such as the shift from agricultural to office-based work or the adoption of the Internet. He even flags possible areas within medicine that are likely to be impacted from customer service to data analysis to patient triage. That prediction lines up with what health systems are already grappling with as they deploy AI across claims processing, prior authorization, and administrative documentation. These disruptions will likely continue into the clinical space once AI adoption for clinical use cases expands.
Dual-Use Risk Is a Health Care Cybersecurity Problem, Today
Gates’s second concern is that AI is empowering bad actors faster than defenders can keep up, particularly against critical infrastructure like hospitals and financial systems. This is not a hypothetical for the sector. We discussed this dynamic in detail following the announcement of Project Glasswing, an initiative built around AI-powered cyber defense for health care, financial services, and other critical infrastructure. The same underlying tension Gates describes, that AI capable of finding a vulnerability can just as easily help exploit one, is precisely why health care organizations cannot treat AI security as a subset of general IT risk management. It requires its own governance track, tied directly into incident response planning, and compliance with long-standing and emerging security and data protection laws at the federal, state, and international levels.
The Psychosocial Risk to Minors Has Already Reached the Courtroom and the Medical Board
Gates spends a substantial portion of his essay on AI companions and their potential to displace human relationships, particularly for young people, citing research suggesting heavier and more emotionally personal chatbot use correlates with worse outcomes. As we discussed previously, lawsuits filed in 2025 alleged that AI chatbots encouraged the suicides of minors. Just this May, we also discussed Pennsylvania Board of Medicine’s suit alleging that an AI chatbot engaged in the unlicensed practice of medicine by holding itself out as a psychiatrist.
State legislatures have moved quickly in response. Colorado, along with more than two dozen other states, has enacted AI legislation touching health care, and companion-AI-specific statutes have emerged that impose transparency, safety, and accountability obligations aimed squarely at protecting minors and other vulnerable users. Although Gates’s essay treats this as an emerging concern, for health care counsel that has been following granular developments, it is already an active compliance and litigation risk category with its own developing regulatory landscape and case law.
To help clients navigate this evolving patchwork of laws, our firm recently launched an interactive State AI Law Tracker, covering enacted AI legislation across health care, life sciences, and employment. Follow more AI regulatory developments using Epstein Becker Green's State AI Law Tracker.
Why Gates’s View Indicates AI Governance Is Key
Gates’s essay is most prescient, not in any individual risk explicitly noted, but in his observation that while AI governance is central to long-term success, no single institution (public or private) currently has visibility across the full AI risk or regulatory picture. He goes on to propose a new institutional architecture built to see across these silos as a means to achieve more thoughtful oversight and regulation. So what are health care and life sciences organizations to do while they wait for this grand governance vision to manifest?
Interestingly, health care and life sciences organizations already face a version of this same structural problem that Gates describes, and are well-positioned to address it on a hyper-local basis. AI governance in health care and life sciences organizations today frequently sits split across compliance, IT security, HR, and clinical leadership, each with visibility into one slice of the risk. Organizations need not wait for a new global governance institution to be formed, but rather can address this issue in hyper-local ways that are right-sized to their organizations. Such a hyper-local approach is more feasible now and worth pursuing now to reduce the risk of enforcement actions or lawsuits that are arising in real time.
It is also important to note that the governance question is clearly one that goes well beyond whether the technology works. It is whether the organization has governance in place to manage the myriad risks it faces across workforce, cybersecurity, operations, supply chain, regulatory compliance, privacy, intellectual property, and many other areas bearing possible legal and financial consequences. Clients standing up AI governance committees are increasingly recognizing the wide scope of areas requiring oversight and assigning ownership to evaluate and manage these risks.
Here are some high-level thoughts on AI governance that may serve as a starting point to foster the work that ought to be done today.
* * * *
If you have questions, please reach out to the author(s).
The Health Law Advisor blog is currently edited by Emily Chi Fogler.
Authors
Member of the Firm