The changing medicolegal landscape: AI in healthcare and its implications for professional indemnity insurance

Key takeaway

Insurers, brokers, healthcare organisations and healthcare professionals should carefully assess AI-related exposures and consider whether existing policy wordings adequately respond to these AI evolving risks in the healthcare sector.

Overview

Artificial intelligence (AI) has the potential to transform healthcare, with applications ranging from diagnostic imaging and predictive analytics to clinical decision-support systems and autonomous robotic surgery. These technologies have the potential to improve patient outcomes, enhance efficiency and support earlier diagnosis, but they also introduce new medico-legal and insurance challenges.

As AI becomes more integrated into healthcare delivery and is expected to play a greater role in clinical decision making, traditional lines of accountability will become less clear. Determining responsibility for adverse health outcomes, may no longer be straightforward, with potential liability extending beyond the treating practitioners and healthcare organisations to AI developers/vendors and medical device manufacturers with integrated AI for AI related errors and omissions.

Against a potentially changing medico-legal landscape, insurers and healthcare providers will need to carefully consider whether existing risk management frameworks and insurance arrangements are equipped to respond to these emerging exposures.

AI in clinical practice: opportunity and risk

AI promises tangible benefits across healthcare. Applications such as early sepsis detection and AI-assisted breast cancer screening can help clinicians identify disease earlier, improve diagnostic accuracy and support more timely interventions. As these technologies continue to evolve, they offer the potential to reduce diagnostic errors, accelerate treatment, and ultimately save lives.

However, the use of AI in healthcare diagnosis and decision making also raises important medico-legal considerations. From a liability perspective, one of the challenges is that the reasoning underpinning AI generated outputs may not always be transparent or readily capable of explanation. In some cases, clinicians may receive a recommendation or prediction without being able to fully understand how the system arrived at that outcome. This lack of explainability can make it difficult to assess the reliability of AI generated information, particularly where it influences diagnosis, treatment recommendations or patient management decisions. It also raises important questions regarding accountability when adverse outcomes occur.

The use of AI in healthcare has the potential to reshape traditional concepts of professional responsibility. For healthcare professionals using AI tools to support diagnosis and treatment, the focus is likely to shift from whether reasonable skill and care was exercised to whether reliance on AI generated recommendations was appropriate and supported by adequate clinical oversight. How these issues will be measured against established standards of professional competence will present as an emerging challenge for courts, regulators and insurers alike.

The regulatory landscape

At the regulatory practitioner level, the Australian Health Practitioner Regulation Agency (AHPRA) has published dedicated guidance on AI in healthcare, signalling that regulators are actively scrutinising how practitioners engage with these technologies. AHPRA's position is that existing regulatory frameworks, including codes of conduct, professional standards, and the National Law, already apply to the use of AI, and that no "regulatory gap" excuses practitioners from their obligations when deploying AI tools. AHPRA has indicated it will consider AI related conduct in the context of notifications (complaints) about practitioners, meaning that inappropriate reliance on AI, failure to verify AI outputs or inadequate disclosure to patients could form the basis of regulatory action.

Liability in healthcare: who is responsible when AI gets it wrong?

AI assisted healthcare challenges conventional assumptions about where responsibility sits when patient harm occurs. An adverse outcome may involve a clinician’s reliance on AI generated information, the health organisation's decision to deploy the technology or the design and performance of the AI system itself. This creates a more complex liability pathway than traditional clinical negligence claims, particularly where the alleged error arises from:

• an opaque algorithmic process of the AI software or embedded medical device;
• from the interaction between human judgement and machine-generated recommendations; or
• failures in the design, functionality or performance of the AI system/medical device.

In Australia, these issues remain largely untested in the clinical context. At present, AI is generally used to support clinical decision making rather than operate as an autonomous decision maker. Healthcare practitioners therefore remain responsible for exercising independent professional judgement, appropriately interrogating AI outputs and recording the clinical reasoning that informs their decisions. As AI tools become more sophisticated and more deeply embedded in healthcare pathways, the way existing negligence principles apply to AI-assisted healthcare is likely to attract increasing scrutiny from regulators, insurers and the courts.

Ultimately, under the negligence framework in Australia, a non-delegable duty of care exists, meaning the final judgment and all patient care decisions rest with the medical practitioner. AHPRA has reinforced this position, making clear that the use of AI does not change a practitioner's existing professional obligations under the Health Practitioner Regulation National Law. Registered health practitioners remain responsible and accountable for their decisions regardless of whether AI tools were used in reaching them. AHPRA expects practitioners to understand the AI tools they use, including their limitations and potential for error, and to exercise professional judgment rather than uncritically accepting AI generated outputs.

Practitioners must also consider their informed consent and patient privacy obligations and patients should be made aware where AI has played a material role in their diagnosis or treatment planning.

Implications for professional indemnity insurance

The emergence of AI in healthcare presents significant challenges and opportunities for the professional indemnity insurance market. As healthcare providers are predicted to increasingly rely on AI-generated information to inform clinical decisions, insurers will no doubt seek assurance that appropriate governance, human oversight and verification processes are in place.

We are also likely to see changes to policy wordings. AI exclusions may emerge in professional indemnity and medical indemnity policies to address claims arising from AI-generated content or autonomous system outputs. At the same time, there are early indicators that insurers may have an appetite to develop dedicated AI products or endorsements to cover AI related risks, that fall outside traditional professional indemnity cover.

A related challenge is the potential for uncertainty regarding how AI related losses are allocated across traditional insurance products and the potential coverage gap between professional indemnity, product liability and cyber insurance. Traditional professional indemnity policies are designed to respond to claims arising from human professional judgement, while cyber policies focus on data breaches and technology failures and product liability on loss arising from medical device failures. As AI becomes embedded in diagnostics, clinical decision support and healthcare administration, there is a risk that AI driven errors in healthcare may fall between the intended scope of policy coverage, creating a potential gap in cover.

As regulators increasingly scrutinise the use of AI in healthcare, questions may emerge as to whether investigations into a health practitioner's professional practice and the use of AI assisted clinical decision making, will trigger inquiry costs and expenses cover in the same way as traditional professional conduct investigations. The adequacy of existing policy wordings to respond to these evolving regulatory exposures is likely to become an increasingly important consideration for insurers.

Practical considerations for stakeholders

For insurers and brokers

The adoption of AI in healthcare requires insurers and brokers to reconsider how emerging risks are identified and insured. While AI-related claims are yet to be tested in Australian courts, insurers should be evaluating whether existing professional indemnity and medical indemnity policies adequately respond (or exclude) to claims arising from AI-assisted clinical decision-making, regulatory investigations and autonomous system outputs.

Particular attention should be given to potential coverage gaps between professional indemnity and cyber insurance, including whether AI-related claims may fall outside the intended scope of either policy. Insurers should also consider whether policy wordings require amendment to address AI specific exposures and whether the market will demand dedicated AI endorsements or standalone AI products as adoption increases. From an underwriting perspective, healthcare organisations' AI governance frameworks, oversight mechanisms and verification processes are likely to become increasingly important underwriting consideration.

For healthcare organisations

Healthcare organisations should ensure that the adoption of AI in clinical settings is underpinned by robust governance and risk management frameworks. As AI is expected to become increasingly integrated into diagnostics, supporting clinical decision making and treatment pathways, healthcare organisations should establish clear policies governing AI use, maintain appropriate audit processes and ensure meaningful human oversight of AI assisted clinical decisions occurs.

Contractual risk allocation will also become increasingly important. As healthcare organisations utilise AI software and procure TGA approved AI embedded medical devices, careful consideration should be given to contractual warranties, indemnities and liability provisions in commercial agreements for these products. Particular attention should be paid to whether responsibility for AI related errors, omissions or system failures can be appropriately allocated to developers or technology providers, where those risks arise from failures or errors in the design, functionality or performance of the AI system.

Healthcare organisations should also review their insurance arrangements to ensure emerging AI related exposures are adequately addressed. This includes assessing whether existing professional indemnity, cyber and product liability policies respond appropriately to AI related claims, regulatory investigations and disputes involving AI vendors and technology providers. Given the potential for AI related losses to fall between traditional insurance policy products, healthcare organisations should identify and address any AI coverage gaps.

For healthcare practitioners

For healthcare practitioners, the key message from AHPRA is clear: the use of AI does not diminish professional responsibility. Practitioners remain accountable for clinical decisions, regardless of whether AI has been used to support them. AI should therefore be viewed as a clinical tool rather than a replacement for professional judgement.

Practitioners should understand the capabilities and limitations of the AI systems they use, exercise independent clinical judgement, verify AI-generated outputs where appropriate and maintain clear documentation of their decision-making processes. They should also remain mindful of informed consent and privacy obligations, where AI plays a material role in diagnosis or treatment planning. As regulatory scrutiny of AI use continues to develop, these practices are likely to be critical in mitigating both liability and regulatory risk.