Will HHS enhance or hinder the promise of artificial intelligence in health care? – Healthcare Blog
Steven Zecora
The U.S. Department of Health and Human Services (“HHS”) acknowledged in its Artificial Intelligence Strategy (V.3): “For too long, our department has been mired in bureaucracy and busywork.” HHS pledged to accelerate artificial intelligence (“AI”) innovation, including “accelerating FDA drug and biologics approvals.”
History shows that well-intentioned but cumulative regulatory interference—more important than scientific complexity—is a major obstacle to rapid technological progress. If AI is subject to typical patterns of regulatory creep, its potential to accelerate drug discovery and development will be significantly reduced. To avoid this outcome, HHS should develop a plan premised on a zero-based regulatory approach. That is, every new technology such as artificial intelligence should be started from scratch, and only the minimum requirements necessary to show effectiveness and safety should be applied in the approval process of the technology.
pace of innovation
Medical innovation lags behind other sectors of the economy. As Dr. Scott Podolsky of Harvard Medical School observed: “Medicine in 2020 is closer to medicine in 1970 than medicine in 1970 is to medicine in 1920.” Podolsky noted that breakthrough drugs such as antibiotics, antihypertensives, antidepressants, antipsychotics and steroids have not had the same impact as innovations of the past 50 years.
There are two explanations for this phenomenon: 1) the inherent complexity of biological processes; 2) the regulatory approval process.
As a baseline for comparison in the following case studies, 4G communications have been in development for less than ten years, with discussions starting around 2001, technical specifications released in 2004, and the first commercial network launched in 2009.
Regulatory intervention in new technologies
- Human Genome (Great Science Leads to Regulatory Paralysis)
The Human Genome Project (HGP) lasted from 1990 to 2003 and is hailed as one of the world's greatest scientific achievements. The project identified the specific locations of genes and DNA, created a “roadmap” of the human genetic code, and facilitated the identification of disease-related genes.
The focus of the Human Genome Project is to balance rapid scientific progress with ethical safeguards. Oversight is primarily managed through internal ethics programs and international data sharing agreements, rather than through a single overarching legislation or regulatory agency.
Under this structure, the Human Genome Project is two years ahead of its target date. That said, the complexity of the problem caused no delays, and progress was not hampered by standard drug approval bottlenecks.
However, once the genetic roadmap for drug discovery and development was handed over, progress slowed dramatically.
Nine years later, scientists were able to demonstrate that CRISPR-Cas9 technology can edit DNA. From that point on, it took eleven years for the first gene-editing therapy to be approved. Three more years later, the FDA announced it would implement greater regulatory flexibility for cell and gene therapies, noting that stakeholders were not always fully aware of its process.
Clarity isn't the only problem, or even the main one. CRISPR technology currently has countless applications, ranging from basic molecular research to advanced clinical, agricultural and industrial uses. Although only one CRISPR therapy (Casgevy) has received official FDA approval as of early 2025, there are hundreds of active research projects, clinical trials for diseases such as cancer, and more than 1,000 CRISPR screens for drug discovery. There is a lack of regulatory resources to handle all these activities. Further delays are inevitable.
Looking back, we can see that once the Human Genome Project's drug discovery and development “roadmap” was handed over, costs and development time during the regulatory review process increased dramatically. The delay appears to be due more to the FDA's lack of resources to manage its own regulations than to the complexity of the scientific issue.
- Stem cells (regulatory compliance ignores financial constraints)
Geron began a human embryonic stem cell research program in the late 1990s, becoming a pioneer in the field and raising $100 million to conduct research and pursue major treatments. Under the guidance of the FDA, Geron conducted more than 2,000 experiments on mice and rats and spent a significant portion of the funding on preclinical research. Geron's Investigational New Drug Application consisted of 22,000 pages and cost $45 million. The company received approval in 2009 to move forward with Phase 1 clinical trials. There were no safety concerns among the initial trial participants, but Geron ended the project in 2011 for financial reasons.
Research on stem cells continues to this day, with applications ranging from Parkinson's disease to damage to the optic nerve. Meanwhile, Geron diverted remaining funds into oncology, winning FDA approval for its telomerase inhibitor in 2024 after 34 years of research.
The Geron case is an example of what happens when regulation does not take into account the time required to enter a market or the cost of entering a market. In other words, given the financial constraints of research, regulatory delays often indirectly hinder innovation. In this case, regulatory intervention also completely changed the direction of the research program.
- Personal genetics (over-regulation denies innovative medical information)
Direct-to-consumer genetics provides an example of information itself becoming a regulated product. Founded in 2006, 23andme is a pioneer in consumer access to genetic insights. In 2013, the FDA determined that providing personal genetic information is a “device” “intended for use in the diagnosis of disease or other conditions, or for the cure, mitigation, treatment, or prevention of disease” and that in some cases “it may result in patients receiving preventive surgery, chemoprophylaxis, enhanced screening, or other disease-inducing actions.” Therefore, the FDA believes that 23andMe violated the Federal Food, Drug, and Cosmetic Act by marketing its products without prior FDA approval.
The FDA noted that 23andMe provides individual reports for 254 diseases and conditions, and the FDA anticipates that each of 23andMe's reports will receive prior approval. As 23andMe CEO Anne Wojcicki explained in response, the FDA system would require over a million tests, which would be virtually impossible for either party to adapt to.
The FDA first approved 23andme in 2015 and continued to authorize various direct-to-consumer marketing reports over the next several years. However, regulation changed the company's direction, and 23andme declared bankruptcy in March 2025.
In this case, regulators extended their tentacles to information services for paternalistic reasons. That said, the FDA doesn't trust users with their data. This slows down the pace of innovation and reduces the value of information.
These three cases illustrate how years of scientific research can turn into decades of regulatory scrutiny and delays. Private equity and public equity markets are closely watching the regulation of artificial intelligence in healthcare to determine how long the first round of AI-derived solutions will be delayed.
FDA’s Current Regulation of Artificial Intelligence – Will History Repeat Itself?
The FDA requires every drug to meet “reasonable assurance of safety and effectiveness,” as set forth in the 1962 regulations. According to FDA procedures, drugs discovered by AI must also comply with the same conventional Phase I, II and III clinical trials as all drug candidates.
Additionally, in January 2025, the FDA launched a 7-step framework for AI submissions. The framework requires sponsors to define specific “contexts of use,” assess risk levels, and establish trust in AI outputs before they are used in regulatory decisions.
The FDA also requires developers to provide detailed documentation about algorithm architecture, training data and potential biases, which is not required for traditional drug development.
In December 2025, HHS issued a request for information regarding the use of artificial intelligence in patient care. As identified in preliminary comments in the proceeding, practitioners are concerned about liability in situations where AI technology is used in clinical care and FDA subsequently finds that prior approval is required.
As the market embraces the use of artificial intelligence in clinical care, it's unclear where the U.S. Department of Health and Human Services will go. However, given that AI innovation is still in its infancy, the U.S. Department of Health and Human Services must select winners to accelerate the adoption of AI in clinical care. This will slow market momentum and increase uncertainty. Even without additional regulation, this uncertainty itself can slow innovation.
Zero-based supervision approach
A zero-based regulatory approach will end regulatory creep and focus regulators on key requirements to demonstrate effectiveness and safety. Regulatory agencies will not be involved in preclinical studies. Documentary evidence will be limited to meeting specific requirements to demonstrate safety and effectiveness.
In this approach, the government's role would be to act as auditor to verify the results of long-term trials. This function will include experimental validation, mechanistic understanding, and ethical oversight.
This approach would free up FDA resources so that more trials can be conducted while still subject to necessary oversight. With experience (and the application of artificial intelligence), applicants and regulators can fine-tune requirements during annual zero-based regulatory reviews.
This approach is not deregulation. It is precise adjustment. Artificial intelligence programs can make audits faster, more relevant, and more accurate. Real-time data can be provided to the FDA, with automated auditing and exception reporting of key variables.
in conclusion
In efficient markets, leaders will leverage AI to accelerate innovative solutions, simplify development, and deliver high-value experiences. The healthcare industry's greatest threat to this outcome is regulation, not disease complexity.
HHS leaders should ask staff to start from scratch with an AI regulatory plan, proposing regulatory requirements only where safety and effectiveness need to be demonstrated. This does not require legislative approval or HHS rule changes.
The FDA’s role will shift from that of an overzealous gatekeeper at every step of the regulatory process to that of a real-time auditor of scientific progress and the innovative solutions that result. This approach will enable staff to observe and evaluate more innovative developments. We know that innovation begets more innovation, which will get the healthcare industry back on track.
Artificial intelligence is in its infancy in healthcare. HHS should treat its regulations in this context. As it stands, the current FDA regulatory framework will not accelerate AI innovation in the healthcare industry and should be adjusted before the initial timeline is missed.
Steven Zecola is a former technology executive and government official. He retired 24 years ago after being diagnosed with Parkinson's disease. He currently works as a passionate patient advocate.
