AI rewrites the rules

Parent Page Home
Parent Page News
Parent Page Digital Health & Wellness 2026
AI rewrites the rules

In brief

AI is becoming embedded across diagnosis, treatment, patient engagement and healthcare administration.
As AI adoption accelerates, accountability and liability are becoming more difficult to define and manage.
Future success will depend on balancing innovation with robust risk management, governance and patient trust.

The digital health and wellness sector is emerging as the testing ground for how AI operates in high-stakes environments, where errors can directly affect patient outcomes. AI is woven through every layer of digital healthcare. It is powering faster diagnosis, smarter triage, and more personalised care. In addition, administrative workflows are increasingly inclusive of automation across every aspect of documentation including billing, scheduling, and operations.

The accountability problem
What does AI look like in digital health 2026?
Big bets on AI and robotics in 2026
Enabling innovation
Best practices and risk controls

The accountability problem

While humans are still responsible for legal oversight, AI is making accountability harder to assign, as the distinction between human and machine-delivered care is becoming increasingly blurred. This is particularly the case when it may be unclear whether patients understand how and when they are talking to a bot versus a clinician, and when the record of these interactions is unclear, for example where the algorithm fails to transfer critical information.

“Administrative AI, including triage, intake, chatbots and portals is already affecting claims, as patients may not realise, or consent to the fact, that they are interacting with chatbots. This increasing complexity in the way that information is received and provided is making it easier for patients to allege miscommunication.”

Keri Marmorek, Claims Team Leader, Beazley
As treatment choices proliferate, a new, more fundamental question is emerging over how to determine what form of treatment is in the patient’s best interest.

“Organisations are navigating a live question of whether the use or non-use of AI could affect the standard of care, potentially exposing them to liability either for not adopting new technology or for failures when using it.”

Carolyn Conners, Head of Healthcare, Beazley

Despite the emerging tension over care models and standard of care, organisations are doubling down on the increased use of AI and robotics in their operations.

Big bets on AI and robotics in 2026

Use of AI is set to increase over the next 12 months:

76% agree they are planning to create generative AI systems for use in their business
81% are planning to use or increase the use of generative AI in the treatment of patients
77% are planning to use or increase the use of AI in the diagnosis of patients
78% are planning to use or increase the use of robotics to create, provide or support new healthcare services
79% agree that the future of healthcare is personalised services driven by greater data and tech insight

This represents a significant shift from 2024 when concern over such issues meant the focus was on bricks and mortar expansion. In fact, developing a face-to-face service dropped from 1st to 7th priority between 2024 and 2026.

Although 79% of executives express confidence in their AI risk management capabilities, confidence in managing bias and trust-related risks is growing. However, concern surrounding intellectual property (IP) protection remains consistently high since 2024.

74% are concerned about bias in generative AI (78% in 2024)
73% are concerned about trustworthiness of AI (80% in 2024)
79% are concerned about intellectual property protection (same as 2024)

Source: Beazley’s Digital Health and Wellness surveys undertaken in 2024 and 2026 with 600 executives from digital health and wellness businesses based in Europe, the US, Canada and Asia.

Such concerns should not be taken lightly.

Trust is essential yet fragile in virtual care services. Patients need confidence that their data is protected and systems are reliable. Clinicians must trust AI-generated inputs enough to use them in practice. Bias is equally critical, as under-represented groups in training data can face mis-triage or misdiagnosis, create clinical safety risks and expose organisations to regulatory scrutiny.

Enabling innovation

As AI makes new approaches possible, insurance is being called upon to aid innovation.

As we look to the future, digital healthcare companies will increasingly need to rely on experienced insurance partners with the knowledge, insight and data to risk-share with businesses.

“Right now, we are playing in a sandbox of AI exposures and possibilities. We are actively underwriting through emerging risks generated by the use of AI, including treatment without contemporaneous human oversight. We can take that risk and be part of this innovation journey.”

Evan Smith, Growth Leader Global Healthcare, Beazley

Best practices and risk controls

As AI becomes embedded, the insurance market is looking hard at what constitutes best practice for insureds. This includes the importance of data sourcing (focusing particularly on concern over copyrighted or sensitive information) and the role of supplemental questionnaires in identifying and managing AI-related risks.
"We’re revising our healthcare professional services, tech and cyber
coverages and developing affirmative AI coverage across our coverages. This includes explicitly addressing AI use in diagnosis, prescribing, treatment decisions and cyber and tech exposures. We have introduced a supplemental AI questionnaire and are developing AI underwriting guidelines and affirmative policy language to provide greater clarity around coverage intent and scope."

Marc Martin, Product Leader Digital Healthcare, Beazley

The accountability problem

The accountability problem

While humans are still responsible for legal oversight, AI is making accountability harder to assign, as the distinction between human and machine-delivered care is becoming increasingly blurred. This is particularly the case when it may be unclear whether patients understand how and when they are talking to a bot versus a clinician, and when the record of these interactions is unclear, for example where the algorithm fails to transfer critical information.

“Administrative AI, including triage, intake, chatbots and portals is already affecting claims, as patients may not realise, or consent to the fact, that they are interacting with chatbots. This increasing complexity in the way that information is received and provided is making it easier for patients to allege miscommunication.”

Keri Marmorek, Claims Team Leader, Beazley
As treatment choices proliferate, a new, more fundamental question is emerging over how to determine what form of treatment is in the patient’s best interest.

“Organisations are navigating a live question of whether the use or non-use of AI could affect the standard of care, potentially exposing them to liability either for not adopting new technology or for failures when using it.”

Carolyn Conners, Head of Healthcare, Beazley

Despite the emerging tension over care models and standard of care, organisations are doubling down on the increased use of AI and robotics in their operations.

What does AI look like in digital health 2026?