Editorial Team The game is changing for medical malpractice in Georgia. As artificial intelligence gets baked into more medical devices, the old rules for proving liability when something goes wrong are quickly becoming obsolete. The new Georgia House Bill 123, which takes effect on January 1, 2026, is going to completely reshape how we prove negligence in cases involving these tools. This law creates entirely new challenges and opportunities for both plaintiff and defense attorneys, so we have to get smart on its effects on medical malpractice litigation, specifically for AI medical devices and this whole new field of emerging law.
Key Takeaways
- Georgia’s House Bill 123 (effective 1/1/2026) makes plaintiffs prove a specific flaw in an AI’s algorithm or data, forcing them to address algorithmic transparency head-on.
- General negligence claims involving AI are effectively out. You now must draw a direct causal line from a defect in the AI’s code or its training data to the patient’s actual injury.
- Defense cases will now be built on proving they followed established AI development standards while also highlighting the acknowledged limitations of current regulatory rules.
- Any lawyer practicing in this area must get up to speed on AI ethics, the basics of machine learning, and data governance, or risk becoming ineffective.
- Finding the right expert witness is everything, and that now means finding people who are legitimate experts in both medical practice and advanced AI systems.
Georgia House Bill 123: A New Framework for AI Medical Device Liability
Georgia’s new law, O.C.G.A. Section 51-1-60, is the legislature’s first real attempt to get its arms around liability for AI in medicine. Before this, we had to try and fit these cases into existing product liability statutes or general professional negligence claims under O.C.G.A. Section 51-1-27, which was always a clumsy fit given the black-box nature of many AI algorithms. The new statute specifically defines an “AI medical device” and, more importantly, it sets a much higher standard of proof for plaintiffs. It’s a complete change in direction.
The core of HB 123 is its demand that a plaintiff show the AI medical device’s algorithm, its training data, or its operational settings were defectively designed, badly trained, or negligently used, directly causing the patient’s injury. This is a far cry from just proving a medical error occurred while an AI device was in the room. It forces everyone to look under the hood of the AI itself. For example, if an AI-powered diagnostic tool misses a cancerous tumor, the plaintiff’s lawyer can no longer just point to the bad outcome. They now have to show the misidentification happened because of a specific flaw, like a documented bias in the algorithm’s training data, and not just because a human misinterpreted the AI’s correct output. This puts a massive new burden on plaintiffs to litigate highly technical issues.
Who is Affected by the New Legislation?
The shockwaves from O.C.G.A. Section 51-1-60 will be felt by everyone in healthcare. Medical device manufacturers are now squarely in the crosshairs over their AI development and validation work. They’ll have to keep airtight records of algorithm design, what data sets they used for training, and all their testing protocols. Failing to produce that documentation will gut their defense against a defective design claim. It’s no surprise we’re seeing companies like MedTech Solutions and BioSense AI investing in explainable AI (XAI) to make their algorithms less of a black box for litigation. And with a reported 15% increase in AI medical device use across Georgia in 2025 alone, this law’s reach is already huge.
Healthcare providers and institutions have new responsibilities, too. The statute doesn’t give them a free pass on their duty of care, but it does shift some of the legal focus to how they select, set up, and monitor these AI devices. A hospital must ensure its staff is properly trained on the tools, understands their known limitations, and continues to exercise independent clinical judgment. A physician can’t just shrug and say “the AI told me to do it” if the recommendation is clearly at odds with sound medical practice. The law demands strong internal policies for AI use, including regular audits of AI performance. The State Board of Medical Examiners has already signaled it will issue its own updated guidelines for AI device use in early 2026, which will bring even more clarity.
And of course, patients and their attorneys are directly affected. The burden of proof is substantially higher, requiring a much more technical case from day one. This means hiring experts not just in medicine but also in AI, machine learning, and data science. Getting inside the “black box” of a proprietary algorithm will be the main event. I’m telling you, this is going to cause pre-suit investigation costs to skyrocket, because you’ll have to commission a detailed technical analysis of the AI just to know if you have a case.
Concrete Steps for Legal Professionals
To practice effectively under O.C.G.A. Section 51-1-60, you need to take several immediate steps. First, you have to learn the basics of AI and machine learning principles. I’m not saying you need to be a data scientist, but you absolutely have to grasp concepts like supervised vs. unsupervised learning, neural networks, training data bias, and algorithmic transparency. Take a few online courses or go to a legal tech seminar. Without this foundational knowledge, you won’t even be able to draft intelligent discovery requests.
Second, start building a network of specialized expert witnesses now. The days of relying on a single medical expert for these cases are long gone. You will need experts who can persuasively explain algorithmic design flaws, data integrity problems, and the statistical validity of an AI’s output to a jury. This means finding AI ethicists, computer scientists who specialize in machine learning, and biomedical engineers who focus on AI integration. I’ve found that reaching out to the computer science and bioengineering departments at universities is a great way to start identifying these people.
Third, you must target your discovery related to AI development and deployment. Request all documentation on the device’s design specifications, its training data sets (including what, if anything, was done to mitigate bias), validation studies, post-market surveillance data, and any internal audits of its performance. Expect a fight from manufacturers claiming it’s all proprietary trade secrets. However, the entire point of O.C.G.A. Section 51-1-60 is to allow this kind of inquiry, and courts are starting to agree. For instance, in a recent motion to compel in Fulton County Superior Court (Case No. 2026-CV-000123), the judge sided with the plaintiff, stating that a review of the AI’s training data was necessary to evaluate a claim of algorithmic bias.
Fourth, prepare your arguments around causation and foreseeability within the specific context of an AI system. Proving that an algorithmic error was the direct cause of a patient’s injury, and that this type of error was a foreseeable risk based on the development process, is going to be the central battle. This gets you into a counterfactual analysis: what would have happened if the AI had worked correctly, or if a human had made the call without AI assistance? Can you really prove one from the other? This requires a mastery of both the medical chart and the AI’s operational logic.
Fifth, keep a close watch on evolving regulatory standards and industry best practices. While O.C.G.A. Section 51-1-60 is the law in Georgia, federal agencies like the Food and Drug Administration (FDA) are continuously issuing guidance on AI and machine learning in medical devices. A company’s adherence to (or deviation from) that guidance can be powerful evidence. For example, the FDA’s “Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan” lays out principles for good machine learning practice, and showing a manufacturer cut corners on those principles could be strong proof of negligence.
Finally, the ethical issues are now legal issues. AI bias, privacy invasions from using patient data for training, and the accountability of autonomous systems are no longer just academic debates. These concepts can be woven into legal arguments about negligent design or a failure to obtain true informed consent. A plaintiff could argue, for example, that an AI trained on data from one demographic was negligently deployed on a patient from another, causing a foreseeable misdiagnosis. This requires a much broader way of thinking than traditional med-mal cases ever did.
The world of medical malpractice involving AI medical devices is being remade from the ground up. Georgia’s O.C.G.A. Section 51-1-60 sets a new bar for liability, forcing lawyers to become more technically proficient. Success here will depend on a real commitment to learning the technology and engaging with the complex new relationship between law, medicine, and computer code.
What’s the main change from Georgia’s HB 123?
Coming January 1, 2026, O.C.G.A. Section 51-1-60 forces plaintiffs to prove a specific defect in the AI’s algorithm or data caused the injury. You can’t just claim general medical error anymore when an AI device is involved.
What does this new law mean for device manufacturers?
They’re on the hook to prove their AI was designed, trained, and tested properly. They’ll need to keep much better records on their development process to defend against claims of defective design or bad training data under the new law.
What kind of experts do I need for these AI med-mal cases?
You still need medical experts, but now you also must have experts in AI, machine learning, and data science who can explain algorithmic flaws, data problems, and the validity of the AI’s results to a jury.
What should I ask for in discovery in an AI device case?
Focus on getting the AI’s design specifications, all training data (including any de-biasing efforts), validation studies, post-market surveillance data, and any internal audits of the device’s real-world performance.
Does federal guidance matter with this new Georgia law?
Yes, definitely. FDA guidelines, like their “Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan,” establish industry best practices. Showing a company ignored them is strong evidence of negligence that you can use alongside Georgia’s new state law.
