By Fatima Aliu
@redlineproject
AI disclosure: Perplexity and Chat GPT to research topics and seek sources for the paper. Adobe Firefly generated the image, and Notebook LM created audio and the FAQ. Read more about our AI use on our principles page.
Summary: This essay explores emerging trends in litigation and regulatory issues arising from the use of Artificial Intelligence across the healthcare industry. The post examines the governance in AI deployment at the federal and state levels, the future of AI systems, and innovation for academic researchers and industry and implications are examined.
Evolution of Artificial Intelligence in Healthcare
In Artificial Intelligence and Healthcare, Hiranji, R et al, 2024, noted that Artificial Intelligence began with Alan Turing’s question on the idea of an artificial device, a tool that could compete with human decision-making. It was, however, the gathering of world’s authorities in data science, engineering and mathematics at the Dartmouth Conference in 1956, that propelled the conception of AI. The goal of the conference was to collaborate and partner in the development of a practical application framework of Ai (Hiranj, R, et al, 2024).
The authors, Hiranji, R et al, stated that the use of AI in healthcare dates to the 1960s and 1970s, when systems were introduced for data analysis and assistive diagnosis. In 1971, INTERNIST-1, the world first artificial medical consultant, was developed as a diagnostic tool through algorithm search using patients’ symptoms. This was significant as an application in clinical diagnostics, beyond research. The Stanford University Medical Experimental Artificial Intelligence in Medicine (SUMEX-AIM) project led with the introduction of MYCIN in 1973 and was used as an assistive tool in diagnostics and treatment of infectious diseases, using a set of inputs. The INTERNIST-1 was expanded with the creation of DXplain by the University of Massachusetts in the 1980s.
The Keragon team noted in History of Ai in Healthcare (2025) that between 2000 through 2010, with advancements in computational data analysis and deep learning , AI application in healthcare, had expanded significantly and integrated into diagnostics, treatment plans, administrative processes and use as a predictive device, increasing accuracy and minimizing errors.
Specifically, AI systems are enabling speedy processes of large data with precision, with pattern recognition in large datasets, enabling disease diagnostics, integration with clinical practices and predictive tools for disease management. Hiranj, R et al, 2024, noted that the introduction of Watson application by IBM in 2007, expanded diagnostics capabilities from input/output to more complex outputs that identify other elements. Pharmbot was developed in 2015 to educate patients on medication and treatment processes. The Kerogen team concluded that Ai systems are “… now integral to the healthcare industry, continually advancing and assisting with improving clinical practices and patient outcomes.”
As of the 2020s, AI systems have advanced healthcare into early disease detection and diagnosis, enabling precision or individualized medicine as an option for patience. Machine Learning and algorithms are used for predictions, and its applications have been in the areas of radiological image and dermatological image interpretations for clinical diagnostics. Other benefits include AI-powered robotic surgery assistance, patients’ disease management, and drug discovery processes through efficacy prediction (Keragon team, 2025). Another study by Hiranj, R et al, 2024, noted that AI systems have improved clinical practices thus improving patients care and outcomes. An article in Artificial Intelligence and Healthcare stated that Ai enables access to health care for patients.
—-
Podcast: Listen to a discussion of the paper topic
—-
Ethical Concerns Arising From AI Application in Healthcare
Weiner, E, et al, 2025, identified five ethical concerns arising from the integration of AI in healthcare. Ethical issues range from justice and fairness, transparency, patient consent and confidentiality, accountability, to patient-centered and equitable care.
Authors noted that there are biases embedded in data algorithms that do not represent all groups or consider differences in healthcare needs, access, usability and cost. The biases can often lead to discrimination and marginalization. An example will be an algorithm that uses healthcare costs as proxies for medical need and the equation of health status of races without accounting for variations. In such cases, care for patients of color are likely to be under-prioritized despite greater health needs.
Authors went further to state that biases can be introduced when there is imbalance and lack of representative training of datasets. Secondly, they noted that many AI models are complex, not easy to understand and lack accountability and transparency. When physicians rely heavily on built in automation biases output without further examination it can further complicate accountability. AI Systems Design for Healthcare Delivery explores this further.
The authors also discussed patient consent and privacy concerns as Ai relies on patients’ sensitive information. They observed that data breaches, misuse, and inadequate oversight in data transfer are some of the ongoing risks. They stated that informed consent, privacy, encryption and anonymization of data, adherence to HIPPA and GDPR are of high priority and must be protected.
They argue for a robust ethical implementation and regulation framework and in their words “to guide the responsible development, validation, and deployment of AI in healthcare. Existing regulations vary by jurisdiction and currently are largely post-hoc. Calls for interdisciplinary collaboration, continuous ethical scrutiny, and inclusive oversight aim to ensure AI benefits all populations equitably and safely. ” Ethical issues arising from AI and healthcare explores this further.
Trends in AI-in Healthcare Litigation Cases
AI deployment in healthcare and emerging litigation areas are identified and classified into five groups. According to the O’Neil’s Institute’s Healthcare Litigation Tracker, the five areas of trending cases include malpractice, consumer/class actions, and contract denial-of benefits lawsuits; Privacy, data-sharing, and consumer protection; algorithmic bias and discrimination suits (and regulatory attention); IP and commercial disputes in genomics and diagnostics — Guardant sues Tempus AI over DNA-testing patents and Regulatory tightening and new compliance obligations.
Brandon Broderick, in his article Medical Malpractice in 2025, outlines composition of medical malpractice in 2025: a misdiagnosis provided by an AI tool or algorithm, Delays in care caused by reliance on automated decision systems, treatment plans developed based on flawed data interpretation and failure by providers to question or override AI-generated suggestions. He noted a 14% increase in AI-related medical malpractice from 2022 to 2024, primarily, in diagnostics in radiology, oncology and cardiology. Disputes over who bears responsibility for errors, between physician and developer, is another area of contention.
On class action claims, lawsuits are being filed against health insurance companies that use AI to deny certain claims under Medicare Advantage are also growing (Estate of Gene B. Lokken et al. v. UnitedHealth Group, Inc. et al., 2023; Barrows et al v. Humana, Inc., 2023; Kisting-Leung et al. v. Cigna Corporation et al., 2023).
The O’Neil Institute notes that the deployment of use of AI in the healthcare sector will continue to have implications for access, affordability and equity. Allegations of biases and discrimination of AI use on claims-processes and diagnostics, are on the rise. Courts are looking to see if AI models were based on representative dataset and did not disadvantage protected groups.
In the case of privacy, data-sharing and consumer protection — Canter, L et al, 2025, in their Data breach and violation article, revealed that Flo Health Inc (Women health app developer) and Google proposed to settle a class action privacy lawsuit for $56 million. Plaintiffs alleged that Flo app improperly shared confidential health data with Google without informed consent. Google settled prior to claims and agreed to pay $48 million. Flo Health Inc. would pay $8 million to the plaintiffs.
AI in Healthcare Governance
The AI in healthcare and life sciences looks to strike a balance with innovation and patients protection. It does not want to over and stunt innovation or under-regulate and bring harm to patients. AI Regulations European Union. The European Union AI Act of 2024 provides the first global framework. Most AI devices used in healthcare classified as high-risk. Product liability directives is stricter.
If an AI-driven system erroneously recommends that causes harm to a patient, the patient can bring a claim. Privacy law and health data sensitivity are protected. The European Health Data Space regulation looks to improve health data access for innovation and training. Accompanying this space will be new technical rules and access conditions.
In the US, AI regulations do not exist at the federal level. The Food and Drug administration (FDA) regulates Ai as a medical and software device by providing guidance. The states and local governments introduced several AI — related bills in 2025. Colorado Artificial Intelligence Act is said to be the first comprehensive AI law in the US and Texas Responsible AI governance Act, both look at High-risk Artificial Intelligence systems that are impactful on the provisions and cost of healthcare decisions.
The state takes a risk-based approach. Utah targets suppliers of mental health chatbots, Virginia targets health facilities that use “intelligent personal assistance.” The United Kingdom is said to be taking an innovation approach under the AI Opportunities Act plan. Among the recommendations being made are to make available “high — impact public dataset to AI researchers and developers. A UK Biobank recently provided anonymized medical data of 500,000 people to researchers in China for medical research
AI in Healthcare — The Way Forward
The question of who bears responsibility for errors that lead the malpractice cases, lingers. In Medical Economics, AI and the New Malpractice frontier, October 2025, and Sara Gerke, an associate professor of law at the University of Illinois Urbana-Champaign, stated that the law in its current form, puts the liability burden on hospitals and physicians. In an evidence-based study of Surgeon’s Perspective on Liability, Duffourc found that physicians often bear liability.
Manufacturers/developers have shared responsibility if there is product defect and if physicians adhered to proper product use. The physicians also felt that patients should be informed and consent to the use of Ai, especially if it could impact outcomes for the patient. The study used to use focus groups with 18 U.S. and EU physicians.
The findings were published in Annals of Surgery Open. On the question of how much should be disclosed to patients on the use of AI tools, Gerke maintained that “The current informed consent doctrine does not necessarily impose a duty to disclose the use of AI in most cases,” Gerke said. There are exceptions — if a patient asks directly, or if an AI tool will play a material role in a procedure. Gerke went further to make an ethical argument, “to proportionate transparency: the deeper a system’s role in diagnosis or treatment, the more disclosure and consent should be secured.”
The European Union AI Act of 2024 sets the global standard framework for AI regulations. It stresses education, compliance and ethical use of AI. Healthcare organizations and individual users are considered deployers. The Act promotes AI literacy for patients and their relatives, staff training, data quality, human oversight, transparency, and monitoring, the Act promotes the safe and effective use of AI in clinical practice. Healthcare organizations are encouraged to aid success and implementation of the systems by balancing innovation with patients’ safety.
____
FAQ
Q1: When did the conception of Artificial Intelligence begin?
The conception of Artificial Intelligence was propelled by the Dartmouth Conference in 1956, which gathered world authorities in data science, engineering, and mathematics. This gathering aimed to collaborate and partner in developing a practical application framework for AI. Earlier, AI began with Alan Turing’s question regarding the idea of an artificial device, a tool that could compete with human decision-making.
Q2: When did AI first start being used in healthcare?
The use of AI in healthcare dates back to the 1960s and 1970s, when initial systems were introduced for assistive diagnosis and data analysis.
Q3: What are some examples of early and later AI applications in medicine?
• In 1971, INTERNIST-1 was developed as the world’s first artificial medical consultant, serving as a diagnostic tool that used algorithm search based on patients’ symptoms. This marked a significant application in clinical diagnostics beyond just research.
• The Stanford University Medical Experimental Artificial Intelligence in Medicine (SUMEX-AIM) project led to the introduction of MYCIN in 1973, which was used as an assistive tool in diagnosing and treating infectious diseases based on a set of inputs.
• INTERNISTS-1 was later expanded with the creation of DXplain by the University of Massachusetts in the 1980s.
• In 2007, the introduction of the Watson application by IBM expanded diagnostics capabilities from simple input/output models to more complex outputs that identify other elements.
II. Current Applications and Benefits
Q4: How is AI currently being used in clinical practice?
AI systems are now integral to the healthcare industry. Between 2000 and 2010, AI applications expanded significantly due to advancements in deep learning and computational data analysis. AI has been integrated into diagnostics, treatment plans, administrative processes, and is used as a predictive device, which helps to increase accuracy and minimize errors.
Q5: What are the key benefits of using modern AI systems in healthcare?
Modern AI systems enable speedy processing of large data with precision and facilitate pattern recognition in large datasets, which allows for disease diagnostics, integration with clinical practices, and predictive tools for disease management. Other benefits include:
• Early disease detection and diagnosis.
• Enabling precision or individualized medicine.
• Applications in radiological image and dermatological image interpretations for clinical diagnostics.
• AI-powered robotic surgery assistance.
• Improving patient care and outcomes, including enabling access to healthcare for patients.
• Drug discovery processes through efficacy prediction.
III. Ethical and Legal Concerns
Q6: What are the main ethical concerns arising from the integration of AI in healthcare?
1. Justice and fairness.
2. Transparency.
3. Patient consent and confidentiality.
4. Accountability.
5. Patient-centered and equitable care.
Q7: How do algorithmic biases impact equitable care?
Biases are often embedded in data algorithms that may not represent all population groups or consider differences in healthcare needs, access, usability, and cost. These biases can lead to marginalization and discrimination. For example, an algorithm that uses healthcare costs as proxies for medical need, or one that equates the health status of different races without accounting for variations, could result in patients of color being under-prioritized despite having greater health needs. Biases can also arise from a lack of representative training datasets.
Q8: What are the trending areas of litigation concerning AI in healthcare?
According to the O’Neil’s Institute’s Healthcare Litigation Tracker, litigation is trending in five classified areas:
1. Malpractice, consumer/class actions, and contract denial-of benefits lawsuits.
2. Privacy, data-sharing, and consumer protection.
3. Algorithmic bias and discrimination suits (and regulatory attention).
4. IP and commercial disputes in genomics and diagnostics (such as Guardant suing Tempus AI over DNA-testing patents).
5. Regulatory tightening and new compliance obligations.
Q9: What specifically constitutes medical malpractice involving AI tools?
Medical malpractice in 2025 related to AI includes: a misdiagnosis provided by an AI algorithm or tool; delays in care caused by relying on automated decision systems; treatment plans developed based on flawed data interpretation; and a provider’s failure to question or override AI-generated suggestions. AI-related medical malpractice increased by 14% between 2022 and 2024, predominantly in diagnostics within radiology, cardiology, and oncology.
IV. Liability and Disclosure
Q10: Who is generally liable when an AI error leads to malpractice?
The question of who bears responsibility for errors leading to malpractice lingers. In its current form, the law typically places the liability burden on hospitals and physicians. Evidence suggests physicians often bear the liability. Manufacturers and developers share responsibility only if there is a product defect and if the physicians adhered to proper product use. Disputes regarding whether the developer or the physician bears responsibility for errors is an area of contention.
Q11: Is a physician required to disclose the use of AI tools to a patient?
The current informed consent doctrine does not necessarily impose a duty to disclose the use of AI in most cases. However, exceptions apply if a patient asks directly, or if an AI tool will play a material role in a procedure. From an ethical perspective, there is an argument for “proportionate transparency,” meaning that the deeper a system’s role in diagnosis or treatment, the more disclosure and consent should be secured.
Q12: What is the global standard for regulating AI in healthcare?
The European Union AI Act of 2024 provides the first global framework and sets the global standard for AI regulations. Most AI devices used in healthcare are classified as high risk under this Act. The Act promotes human oversight, continuous monitoring, staff training, data quality, and transparency.
Q13: Does the US have comprehensive federal regulations for AI in healthcare?
No, AI regulations do not exist at the federal level in the US. The Food and Drug Administration (FDA) regulates AI by providing guidance for medical and software devices.
Q14: Are US states regulating AI in healthcare?
Yes, US states and local governments introduced several AI-related bills in 2025. The Colorado Artificial Intelligence Act is noted as the first comprehensive AI law in the US. Both the Colorado Act and the Texas Responsible AI Governance Act focus on “High risk Artificial Intelligence systems” that impact the cost and provision of healthcare decisions, utilizing a risk-based approach. Utah, for instance, targets suppliers of mental health chatbots, while Virginia targets health facilities that use “intelligent personal assistance.”