Expert Witness Journal Issue 68 August 2026 - Flipbook - Page 104
•
automation bias, where professionals defer to
AI recommendations rather than exercising
independent judgment; and
•
failure to independently verify AI-generated
information.
Disputes could involve multiple parties, including
pharmacists, pharmacy owners, and AI providers,
along with insurers or indemni昀椀ers for most or all
of them. Determining where responsibility lies, and
apportioning losses, could become signi昀椀cantly
more complex than in pre-AI clinical negligence
claims.
Impact on clinical negligence claims
Although generative AI can hallucinate a caselaw
precedent, it cannot hallucinate a drug or
prescription, so the pharmacist’s role in dispensing
medication may not be signi昀椀cantly a昀昀ected.
Regulatory and Fitness to Practise risks
A key regulatory risk will be over-reliance on
AI-generated outputs. While AI may assist with
medicines optimisation, prescribing support, and
patient communications, pharmacists must continue
to exercise independent professional judgment. The
GPhC guidance con昀椀rms that decisions a昀昀ecting
patient care should not be delegated entirely to
automated systems.
However, many risks still exist. The Independent
Prescribing Rollout and expanded “Pharmacy First”
scheme are examples of how pharmacy practice is
changing. These are areas where AI could be used
to help support pharmacists, for example when
prescribing, in medication reviews, and when giving
advice to patients.
The GPhC also highlights the importance of
understanding the capabilities and limitations of AI
systems. Pharmacy providers should ensure that sta昀昀
receive appropriate training and are equipped to
identify potential inaccuracies, bias, or errors in AIgenerated recommendations. Locum pharmacists
may face particular pressure if they have limited
time to familiarise themselves with AI-supported
systems and processes at a new location.
As AI becomes increasingly integrated into
pharmacy services, the key issue is unlikely to be
whether it is acceptable in general for AI to be used,
but whether the pharmacist’s reliance on it in a
particular instance was reasonable and defensible.
The Bolam test will still apply, so pharmacists will
generally not be negligent if their actions would
be supported by a responsible body of professional
opinion. The di昀케culty in future may be proving
what a reasonable pharmacist should do, given that
AI in pharmacy is relatively new, and that AI might
be used to support aspects of pharmacy practice
that are also relatively new. Cases such as Ayinde
v London Borough of Haringey suggest that, in
principle, the AI output should be veri昀椀ed, but
expert pharmacy evidence would still be needed to
explain what steps a reasonable pharmacist should
take in the particular circumstances of the case.
The use of AI also requires consideration around
transparency, governance, and data protection.
Organisations implementing AI-enabled services
should ensure that appropriate policies, monitoring
arrangements, and safeguards are in place,
particularly where patient information is processed
or clinical decisions are in昀氀uenced by automated
systems.
Failure to implement appropriate safeguards may
expose pharmacists and pharmacy providers to
regulatory scrutiny. Where patient safety concerns
arise, AI-related errors could also lead to Fitness
to Practise investigations against individual
pharmacists, concerning competence, professional
judgment, or professional misconduct.
Even if expert evidence is available, in such a
swiftly developing area the Bolitho principle may
be particularly important. This provides that, even
where a body of professional opinion supports
a particular approach, courts may still examine
whether that approach is capable of withstanding
logical analysis.
Liability and insurance considerations
The use of AI in pharmacy also complicates the
insurance or indemnity arrangements for claims.
Pharmacy providers should consider whether their
use of AI is adequately re昀氀ected in their existing
medical indemnity arrangements.
Where an AI system is alleged to have contributed
to patient harm, disputes may arise about whether
the harm resulted purely from clinical negligence,
purely technology failure, or a combination of
both. The injured patient could include clinical
negligence and product liability allegations in the
same action. Alternatively, the patient might sue
only the pharmacist, leaving the pharmacist and
their indemni昀椀er to decide whether to face the
di昀케culties and costs risks involved in bringing a
contribution claim against the AI provider.
EXPERT WITNESS JOURNAL
The GPhC requires pharmacists to have “adequate
and appropriate” indemnity. The Insurance Act
2015 imposes a ‘duty of fair presentation’, meaning
that if material information is not presented when
the policy is purchased, insurance cover can be
withdrawn or reduced later on. Therefore, before
purchasing insurance or an indemnity arrangement,
102
AUGUST 2026
