AI scribing in healthcare: Medicolegal risks, consent and indemnity explained

Understanding AI scribing tools in clinical practice  

Technology continues to reshape the way healthcare is delivered, and artificial intelligence is playing an increasingly visible role. Tools such as AI transcription software are attracting attention from clinicians looking to streamline administrative tasks and spend more time with patients. At Medical Protection, we are seeing rising interest in these tools and helping members understand the medicolegal considerations that come with their use. 

AI transcription tools, often described as ambient or AI scribes, use advanced voice technology to capture consultations and generate written records. This can reduce the time spent on note taking. However, their use introduces a number of important professional and legal responsibilities. 

HPCSA guidance on the ethical use of AI 

The Health Professional Council of South Africa (HPCSA)’s Booklet 20 – AI Guidelines on Ethical Use of Artificial Intelligence contains updated guidance which is generally encouraging of the use of AI tools. However, doctors should be aware of some important factors to consider when using AI scribing in their practice.

The revised version notes that AI technologies have “led to notable changes in the delivery of healthcare” and that overall, the use of AI in providing healthcare “should minimize potential data-related harm… promote the equitable delivery of safe quality care and maintain the integrity of the practitioner/patient relationship.”

Clinical responsibility and accuracy of AI generated medical records 

Doctors should always ensure the notes produced by transcription software are an accurate reflection of the consultation and do not omit any potentially relevant information. It is also important that notes are proportionate and do not contain unnecessary information that may impact future patient care. 

The HPCSA guidelines place ultimate responsibility for the use of AI on healthcare professionals, meaning important issues such as data protection and consent should also be considered when using such programmes. 

Consent 

Any AI tools in use in South Africa must comply with Protection of Personal Information Act (POPIA) standards, which state that personal data can only be processed where the subject consents.

Individual practices should ensure they institute a policy regarding the use of AI, including when informed patient consent should be obtained. Medical Protection advise that informed consent should always be sought from patients before using AI tools that require the sharing of their personal data with a third party. This consent can be obtained in various ways, including verbal consent during consultations, or a physical consent form provided prior to consultations etc. The type of consent given should be documented within a patient’s medical record. The HPCSA’s guidance on decision making and consent provides further guidance. 

Critically, where patients decline the use of AI, the HPCSA’s guidance specifically states that patients “may not be disadvantaged or refused access to health services.” 

Data protection 

Before a clinician or practice starts using an AI system, they should also ensure they understand what personal data it uses, how it uses it, where the personal data goes in situations where a third-party is involved in the processing, whether it is retained by the provider of the AI product or re-used in any way, and how the product can support in meeting POPIA and Promotion of Access to Information Act (PAIA) obligations. 

In addition to compliance with existing legislation, the HPCSA’s guidelinesstate that “anonymisation of data does not provide enough protection to a patient’s information when machine-learning algorithms can identify an individual… with as few as three data points.” It states that medical practitioners have a responsibility to understand AI technologies to mitigate risks to confidentiality.

To ensure that the use of AI software is secure and in line with data protection legislation, we would recommend seeking guidance from the Information Regulator.

Indemnity  

Understandably, doctors may have questions about whether they can request assistance should medicolegal issues arise from their use of AI. 

At Medical Protection, we appreciate the benefits that AI can bring as an adjunct to clinical practice and recognise that members will be making use of this emerging technology. Members can request assistance with matters arising from the use of AI systems, provided that these are not a fully autonomous system and a human retains oversight or final decision-making authority.

As such, members can request assistance in the usual way where a clinical negligence claim arises from a member’s use of AI software, provided the issue relates to the member’s own clinical judgement or actions. This assistance would apply where Medical Protection is providing the member with indemnity for clinical negligence claims. 

Medical Protection would not normally provide indemnity for issues relating to the failure of AI software itself - for example, if the software has been incorrectly programmed or developed. 

Members can seek assistance with other medicolegal matters, such as regulatory investigations, arising from the use of AI software as part of their clinical practice, in the usual way. 

Where a doctor is working in a state hospital, indemnity for claims may fall to the hospital or state health department, so doctors may wish to clarify the indemnity situation with them. 

It is sensible for doctors to take care when entering into contracts or agreements with AI suppliers and be cautious about agreeing to indemnify against a claim. 

Summary 

In this rapidly progressing area, doctors should continue to work in a manner consistent with HPCSA, Information Regulator guidance, and relevant legislation.

Doctors should contact Medical Protection or their Medical Defence Organisation if they have any medicolegal concerns around the use of AI.