Expert Witness Journal Issue 66 April 2026 - Flipbook - Page 79
Do the answers to these questions indicate
that English law as it currently stands is able to
address the role that AI may come to play in dayto-day life?
not itself a legal person, and cannot be approached
as if it were one. The question therefore becomes:
how should liability be ascribed for harms caused by
the autonomous actions of AI?
Generally speaking, the UKJT’s position is that, as
a common law framework, the English legal system
can “provide certainty and predictability in the
context of technological innovation”. As such, the
UKJT sees English law as well placed to adapt to the
challenges of using AI, and to modify or expand
existing principles where necessary to account
for unique issues that arise. Moreover, the report
suggests that areas of “true novelty” will be rare; in
most cases, the UKJT suggests, although the factual
background (i.e. the use of AI) may be novel, “the
application of well-established legal principles is
reasonably straightforward”.
This question is further complicated by other aspects
of the way in which AI functions. For example, the
opacity of the underlying models means it is usually
di昀케cult to identify why a particular decision was
made. Often, multiple parties have been responsible
for developing and training an AI tool, meaning it is
hard to identify who is responsible for a particular
issue. In addition, AI tools may be part of complex
supply chains with multiple suppliers, making it
harder to identify which party is liable for any one
loss.
•
Can English law as it currently exists
account for these novel issues?
However, the UKJT also acknowledges that there is
at least a current “perception of legal uncertainty”
in relation to liability for losses arising from the use
of AI. This uncertainty could hold back adoption of
AI, and lead to potentially unnecessary expenditure
on risk mitigation such as the purchase of insurance.
Accordingly, the UKJT considers it is important to
address this potential uncertainty pre-emptively,
rather than waiting for the issues to be addressed
by the courts.
As the UKJT notes, in many cases the relationship
between the users and providers of an AI tool will
be governed by some form of contract, which will
allocate liability between the parties. It is also likely
that contractual relationships will govern upstream
liability allocation (e.g. between AI model developers
and AI tool developers who use those AI models in
their services or products). Where liability is dealt
with by contract, the novel implications of AI will be
limited to questions of causation.
What is “AI” in this context?
However, the Legal Statement examines several
circumstances where AI may give rise to novel
liability issues:
As the Legal Statement acknowledges, there is
no universally agreed de昀椀nition of AI. The UKJT
therefore adopts “technology-agnostic” de昀椀nition
which captures the novel key characteristics of AI.
The UKJT identi昀椀es “autonomy” as the key novel
characteristic of AI, de昀椀ning AI as any “technology
that is autonomous”.
Negligent use of AI
The primary form of non-contractual liability
the UKJT envisages might arise as a result of AI is
tortious liability for negligence. As in all such claims,
a key question will be whether there is a relevant
non-contractual duty that can be said to have been
breached.
The Legal Statement describes autonomous
technologies as ones which operate such that:
the output that results from a given input is
“unpredictable”; there is an “opacity of reasoning” in
relation to the output generated; and there is limited
power for the user to control the tool’s output. It is
these features that the UKJT considers give rise to
particular legal challenges when ascribing liability
for harms caused by the use of AI.
One scenario where there may be a negligence
claim arising from AI use is where a legal person
(the injured party) is harmed by the use of (or
failure to use) AI by another legal person (the AI
user). The Legal Statement suggests that English
law can already account for such a situation by
applying existing principles of negligence: if the
AI user owed the injured party a duty of care then
they may be liable for harm su昀昀ered by the injured
party if it can be shown that their actions fell short
of that duty of care. The fact the harm is caused
by their application of AI tools (or failure to make
use of AI tools where they should have) has no
bearing on this legal analysis. The Legal Statement
argues that “in many cases the addition of AI will
simply be considered a tool of those who exercise
relevant control over it, and can be said to have been
“responsible for its actions””. The UKJT gives the
Why does AI give rise to legal uncertainty?
The UKJT identi昀椀es the basis for at least the
perception of legal uncertainty as being that
“English private law… has never previously needed
to address the capability of autonomy… other than
in humans”.
Historically, liability has been attributable to a legal
person based on either their voluntary actions,
or those of a human agent acting on their behalf.
However, the UKJT starts from the premise that AI is
EXPERT WITNESS JOURNAL
77
APRIL 2026
