Not
every
mistake
is
attributable
to
an
AI
hallucination.
But
once
a
firm
suffers
a
couple
admonishments
from
the
court
—
including
one
that
cost
the
firm
over
$50K
to
make
right
—
its
lawyers
have
to
redouble
their
effort
to
keep
every
filing
above
reproach.
Because
the
moment
anything
looks
off,
opposing
counsel
will
wonder
if
it’s
the
product
of
AI
run
amok
and
lax
oversight.
And
the
court
isn’t
likely
to
give
the
accused
firm
the
benefit
of
the
doubt.
Gordon
Rees
became
a
cautionary
example
for
AI
hallucination
mishaps
after
one
of
its
attorneys
submitted
fabricated
citations
in
a
bankruptcy
case
last
summer.
In
Jackson
Hosp.
&
Clinic
Inc.,
a
judge
issued
an
order
to
show
cause
why
sanctions
shouldn’t
follow
based
upon
a
brief
with
“pervasive
inaccurate,
misleading,
and
fabricated
citations,
quotations,
and
representations
of
legal
authority.”
The
firm
reimbursed
fees
for
lawyers
inconvenienced
by
the
error
and
threw
itself
on
the
mercy
of
the
court
and
assured
everyone
that
it
had updated
its
AI
policies to
include
a
“cite
checking
policy.”
Fast
forward
to
the
present,
and
a
recently
filed
brief
in
Huynh
v.
Redis
Labs,
claims
that
Gordon
Rees
has
again
submitted
a
brief
with
AI
hallucinations.
In
a
motion
to
compel
battle
entering
its
third
round
after
the
defendant
represented
by
Gordon
Rees
failed
to
comply
with
the
first
two
orders
to
compel.
Despite
earning
monetary
sanctions
and
a
warning
that
terminating
sanctions
could
follow,
the
defendant
still
hadn’t
been
deposed
—
canceling
at
the
last
minute
attributed
to
a
medical
emergency
—
and
the
plaintiff,
represented
by
Bach
Mili
LLP,
filed
a
renewed
motion.
Gordon
Rees
submitted
its
opposition
and,
in
reply,
Bach
Mili
claims
that
a
lot
of
the
citations
don’t
add
up:
The
distortions
go
directly
to
the
governing
standard
for
terminating
sanctions.
Defendants
cite
Doppes
v.
Bentley
Motors,
Inc.
to
argue
termination
is
inappropriate
where
a
party
expresses
willingness
to
comply.
(Opp.
at
7.)
Doppes
holds
the
opposite.
The
Court
of
Appeal
concluded
the
trial
court
abused
its
discretion
by
refusing
to
impose
terminating
sanctions
after
repeated
discovery
violations
and
ineffective
lesser
sanctions,
and
directed
entry
of
judgment
as
a
sanction.
(174
Cal.App.4th
967,
992-994.)
Defendants
invoke
Collisson
&
Kaplan
v.
Hartunian
for
the
proposition
that
terminating
sanctions
should
be
used
sparingly,
though
that
phrase
does
not
appear
in
the
opinion
and
the
case
affirmed
striking
the
defendant’s
answer
for
“lawyer
game
playing
at
its
worst.”
(Opp.
at
4;
21
Cal.App.4th
1611,
1617.)
They
attribute
to
Biles
v.
Exxon
Mobil
Corp.
a
definition
of
willfulness
that
does
not
exist.
(Opp.
at
5.)
They
cite
Rail
Services
of
America
v.
State
Comp.
Ins.
Fund
for
language
that
does
not
appear
in
the
decision.
(Id.
at
4.)
They
represent
Huh
v.
Wang
and
Corns
v.
Miller
as
excusing
noncompliance
based
on
medical
emergency,
though
neither
case
stands
for
that
proposition.
These
are
not
debatable
characterizations.
(Id.
at
5.)
They
are
reversals
of
holdings
and
quotations
that
do
not
exist.
The
brief
doesn’t
explicitly
say
these
issues
were
caused
by
AI
research
and
drafting,
but
it
does
include
the
relevant
local
precedent
holding
that,
“an
attorney
has
a
nondelegable
duty
to
personally
read
and
verify
every
cited
case
and
that
reliance
on
hallucinated
authority
renders
a
filing
frivolous
and
sanctionable.”
“Defendants’
opposition
is
sanctionable
on
its
face
because
it
relies
on
false
and
misleading
legal
authority,
misstates
holdings,
and
fabricates
quotations,”
Bach
Mili
states.
“This
is
not
aggressive
advocacy
or
sloppy
research.
It
is
the
submission
of
false
authority
to
the
Court
in
violation
of
counsel’s
most
basic
obligations.”
The
reply
also
calls
out
a
December
3,
2025
order
in
Villalovos-Gutierrez
v.
Pol,
where
Gordon
Rees
earned
another
reprimand
for
AI-hallucinations.
U.S.
Magistrate
Judge
Carolyn
Delaney
wrote
in
that
order:
In
more
than
one
other
instance,
defendant’s
case
citations
do
not
support
the
specific
explanatory
phrase
presented
alongside
the
citation.
Counsel
shall
not
file
or
otherwise
present
to
the
court
any
documents
which
contain
AI-hallucinated
citations
or
fictitious
or
non-existent
legal
citations.
Counsel’s
failure
to
confirm
the
existence
of,
as
well
as
the
accuracy
and
veracity
of
a
case
or
other
legal
citation
created
by
an
AI
tool
or
taken
from
another
indirect
source,
is
a
potential
ground
for
sanctions.
That’s
December.
That’s
several
months
after
Gordon
Rees
suffered
their
very
public
benchslapping
in
August.
That’s
after
the
firm
promised
its
updated
AI
policy
with
stricter
cite
checking.
But
a
cite-checking
policy
only
works
if
people
follow
it,
and
if
these
allegations
are
accurate,
whatever
internal
reforms
the
firm
implemented
either
aren’t
being
followed
or
aren’t
enough.
Also,
according
to
the
reply,
the
partner
on
Villalovos-Gutierrez
was…
the
exact
same
partner
in
this
case.
That
doesn’t
mean
these
cites
are
necessarily
AI-generated.
Taking
careful
liberties
with
the
meaning
of
cases
cases
is
a
time-honored
tradition
that
predates
ChatGPT
by
a
few
centuries.
But
a
firm
that’s
already
been
dinged
for
hallucinations,
and
a
partner
who
has
already
been
separately
dinged
for
hallucinations,
filing
a
brief
with
a
bunch
of
arguably
incorrect
cites.
It
does
not
take
a
machine
learning
scientist
to
think
AI
is
at
least
arguably
involved.
And
that’s
the
real
problem
for
Gordon
Rees
—
or
any
firm
caught
in
a
hallucination
kerfuffle
—
right
now.
Whether
any
specific
citation
was
generated
by
AI
—
indeed,
whether
any
specific
citation
is
even
wrong
as
opposed
to
merely
debatable
—
opposing
counsel
now
has
every
incentive
to
scrutinize
any
citation
out
of
the
firm
with
a
jeweler’s
loupe.
The
damage
of
a
hallucination
incident
spills
over
into
all
of
the
firm’s
litigation
efforts
and
it
will
take
a
long
time
to
repair
that
harm.
Earlier:
Biglaw
Firm
‘Profoundly
Embarrassed’
After
Submitting
Court
Filing
Riddled
With
AI
Hallucinations
Joe
Patrice is
a
senior
editor
at
Above
the
Law
and
co-host
of
Thinking
Like
A
Lawyer.
Feel
free
to email
any
tips,
questions,
or
comments.
Follow
him
on Twitter or
Bluesky
if
you’re
interested
in
law,
politics,
and
a
healthy
dose
of
college
sports
news.
Joe
also
serves
as
a
Managing
Director
at
RPN
Executive
Search.