Ruiz v. Magellan Financial & Insurance Services

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
WO
IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF ARIZONA
Michael L Ruiz,
Plaintiff,
v.
Magellan Financial & Insurance Services,
Defendant.
No. CV-23-02090-PHX-DWL
ORDER
On July 8, 2026, the Court issued an order to show cause (“OSC”) requiring
Plaintiff’s counsel, Elizabeth Tate, to show cause why she should not be sanctioned or
disciplined for including, in Plaintiff’s response to Defendant’s motion for attorneys’ fees,
two false quotations from Ninth Circuit cases. (Doc. 163.)
In her response to the OSC, Ms. Tate “acknowledges that representing the language
as direct quotations from the cases was a mistake, sincerely apologizes to the Court,
counsel, and the parties, and accepts responsibility for the improper quotations.” (Doc.
171 at 1.) Ms. Tate also provides the following explanation for how the false quotations
came to be included:
Ms. Tate utilized Claude AI to help her research and prepare the Response
. . . . After Claude prepared a draft brief incorporating the arguments Ms.
Tate asked it to make, Ms. Tate entered the following query: ‘Now please
make a list of the cited cases for me to review on lexis to make sure I agree
with the way you’ve cited them.’ Claude generated a list of fourteen cases.
Ms. Tate confirmed they were real cases, read all of them, checked whether
quoted language appeared in the cases, and exercised her professional
judgment to assess whether the cases stood for the propositions for which
Case 2:23-cv-02090-DWL Document 172 Filed 07/28/26 Page 1 of 6

- 2 -
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Claude cited them. In several instances, Ms. Tate determined that Claude
had either misquoted or improperly described a case. In all those instances,
except the two at issue here, she adjusted Claude’s suggested text to strike
the misquotations or otherwise revise the language to conform with her
reading of the case. Ms. Tate did not correct misquotations in the
parenthetical descriptions of two cases . . . . She read both cases and
determined that the parentheticals in the Claude output accurately
characterized the propositions for which she wanted to cite the cases. She
also determined that the quoted language did not appear in the cases, but,
unlike with the cases described in the preceding paragraph, she mistakenly
neglected to fix the errors when she finalized the Response.
(Id. at 1-2, citations omitted.)
Ms. Tate contends that her “conduct does not warrant sanctions or discipline,
particularly when contrasted with the conduct of lawyers in similar matters who were not
sanctioned or disciplined for much more pervasive use of fabricated authorities and
improper citations.” (Id. at 3.) Ms. Tate elaborates: “Unlike the lawyers in the other cases,
Ms. Tate sought to use AI in a responsible manner . . . [and] has been honest and forthright
about how the errors happened. She used AI, she intended to use it responsibly, but she
messed up and did not fix two quotation errors. She is not offering excuses and is not
blaming anyone but herself. She is embarrassed by her mistake and does not intend to let
it happen ever again.” (Id. at 6.) “Ms. Tate asks the Court to forego sanctions or discipline”
but “[i]f the Court is inclined to formally discipline Ms. Tate, the discipline should be
limited to either a reprimand or admonition.” (Id. at 7.)
Although the Court appreciates Ms. Tate’s expressions of contrition, her response
to the OSC raises additional questions that require further development. Ms. Tate contends
that her use of AI should be characterized as “responsible” in part because she follows a
practice of confirming the accuracy of any case citations and quotations that are generated
by Claude AI (“Claude”). But it appears that instead of reviewing the Claude-generated
brief in a thorough line-by-line manner, confirming the accuracy of each citation as it arises
during her review of the brief as a whole, Ms. Tate simply asks Claude to list the cases that
Claude used when drafting the brief and then uses that list (rather than the brief) for
Case 2:23-cv-02090-DWL Document 172 Filed 07/28/26 Page 2 of 6

- 3 -
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
purposes of her verification efforts. (Doc. 171-1 at 2 ¶ 5.) As an initial matter, this
approach has an obvious flaw—it does not include a mechanism for verifying the accuracy
of the list generated by Claude. More important, although the Court recognizes (without
any hint of endorsement) that attorneys are increasingly delegating certain aspects of their
lawyering duties to generative AI programs, it does not strike the Court as responsible to
allow an AI program to draft an entire brief and then submit that brief, without significant
attorney reworking, as the attorney’s own work product. Needless to say, Claude is not a
licensed attorney permitted to practice law in the District of Arizona. It is one thing to use
Claude for assistance with, for example, gathering cases for an attorney’s initial review as
part of the prewriting process. Claude may also be useful in proofreading an attorney’s
draft of a brief. But when an attorney delegates the entire drafting process (other than a
hasty cite-check at the end) to a generative AI program, it is hard to see how that approach
could be characterized as a responsible use of AI. See, e.g., State Bar of Arizona, Guidance
For The Use of Generative Artificial Intelligence In The Practice Of Law In Arizona (Apr.
22, 2025) (“Legal professionals must critically review and refine AI-generated legal
research, citations, arguments, and documents before submission. The duty of competence
requires more than just detecting errors—it requires active legal analysis and validation of
generative AI-assisted work product.”).
In a related vein, Ms. Tate’s response to the OSC suggests that the false quotations
identified in the OSC were anomalies—she contends that she simply “messed up” in these
two isolated instances. (Doc. 171 at 6. See also Doc. 171-1 at 2-3 ¶¶ 7-8 [“In several
instances, I determined that Claude had either misquoted or improperly described a case.
In all of those instances, except the two at issue in the [OSC], I adjusted Claude’s suggested
text to strike the misquotations or otherwise revise the language to conform with my
reading of the case. I read both [Saman and Mitchell]. I concluded that the Claude-
generated text in the parentheticals following the case citations accurately characterized
the propositions for which the cases were cited. I also determined that the quoted language
did not appear in the cases, but, unlike with the cases described in the preceding paragraph,
Case 2:23-cv-02090-DWL Document 172 Filed 07/28/26 Page 3 of 6

- 4 -
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
I mistakenly neglected to remove the quotation marks when I finalized the Response.”].)
In light of this explanation, the Court spot-checked some of the other briefs that Ms. Tate
filed in this case, and it appears that at least three other briefs contain similar errors.
For example, Ms. Tate filed a motion in limine to exclude “evidence that Mr. Rice
allegedly ‘did good things’ for Plaintiff . . . , treated Plaintiff well at times, or maintained
a positive relationship with Plaintiff,” arguing that such evidence “constitutes
impermissible character evidence and improper lay opinion.” (Doc. 74 at 2.) In support
of this request, Ms. Tate argued that “[c]ourts routinely exclude such ‘good acts’ evidence
offered to show that a supervisor or employer could not have engaged in discrimination”
and cited one case in support of that proposition: Ansell v. Green Acres Contracting Co.,
347 F.3d 515 (3d Cir. 2003). (Id.) But as Defendant correctly noted in its response, “[t]he
Ansell court painstakingly analyzed the relevant evidence in that case—favorable treatment
of another older worker in an age discrimination case—under Fed. R. Evid. 404” and
“concluded that the evidence was ‘relevant and admitted for a proper purpose,’ and
affirmed the District Court.” (Doc. 85 at 3, quoting Ansell, 347 F.3d at 518.)
Separately, during trial, Ms. Tate filed “Plaintiff’s Brief on Impeachment.” (Doc.
124.) That brief contained the following purported quotation from United States v.
Antonakeas, 255 F.3d 714 (9th Cir. 2001): “Rule 607 permits impeachment by
contradiction, or the admission of extrinsic evidence to impeach specific errors or
falsehoods in a witness’s testimony.” (Id. at 2.) The quoted language, however, does not
appear in Antonakeas. Elsewhere in the same brief, Ms. Tate provided the following
purported quotation from United States v. Osazuwa, 564 F.3d 1169 (9th Cir. 2009): “[A]
party can introduce otherwise inadmissible evidence when the defendant opens the door by
introducing potentially misleading testimony.” (Id.) But the quoted language, once again,
does not appear in Osazuwa.1
1 Both cases do contain passages that broadly support the point Ms. Tate was
attempting to make. For example, Antonakeas contains the following passage: “The
government responds that Appellant, on direct examination during the defense case-in-
chief, opened the door for rebuttal by making sweeping denials of any involvement in
drugs. We agree that this was proper impeachment by contradiction. Federal Rule of
Evidence 607 allows the admission of extrinsic evidence to impeach specific errors or
Case 2:23-cv-02090-DWL Document 172 Filed 07/28/26 Page 4 of 6

- 5 -
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
And on July 3, 2026, during the post-trial proceedings, Ms. Tate filed a response in
opposition to Defendant’s motion for an extension of time. (Doc. 160.) That brief included
the following statutory citation: “See 28 U.S.C. § 1924 (‘[T]he bill of costs shall be verified
by the affidavit of the party or his duly authorized attorney.’).” (Id. at 2.) In fact, the
statutory language provides as follows: “Before any bill of costs is taxed, the party claiming
any item of cost or disbursement shall attach thereto an affidavit, made by himself or by
his duly authorized attorney or agent having knowledge of the facts, that such item is
correct and has been necessarily incurred in the case and that the services for which fees
have been charged were actually and necessarily performed.” 28 U.S.C. § 1924. Five days
later, on July 8, 2026, the Court issued the OSC pertaining to the two false quotations in
the filing at Doc. 155. (Doc. 163.) The following day, on July 9, 2026, Ms. Tate filed a
“notice of errata” concerning the filing at Doc. 160. (Doc. 165.) In that notice, Ms. Tate
stated that “[t]he quote [concerning 28 U.S.C. § 1924] is incomplete.” (Id. at 1.) But this
benign characterization is inaccurate. Ms. Tate’s purported quotation of § 1924’s statutory
language in the filing at Doc. 160 was not “incomplete”—it was a fabricated paraphrase of
the statutory language masquerading as a direct quotation, similar to the false quotations
that appeared in other filings.
It appears to the Court that the additional examples identified above may also be
attributable to the misuse of generative AI. If so, they are difficult to reconcile with the
notion that the false quotations appearing in Ms. Tate’s filing at Doc. 155 were an isolated
mistake stemming from the otherwise responsible use of AI. Nevertheless, because Ms.
Tate has not had a chance to address these additional examples, the Court will afford her
an opportunity to do so before deciding how to proceed.
falsehoods in a witness’s testimony on direct examination.” 255 F.3d at 724. Meanwhile,
Osazuwa contains the following passage: “In a criminal prosecution, the government may
introduce otherwise inadmissible evidence when the defendant ‘opens the door’ by
introducing potentially misleading testimony. A defendant may open the door by
minimizing, or attempting to explain away, a prior conviction. If a defendant opens the
door, the prosecution may ‘introduce evidence on the same issue to rebut any false
impression that might have resulted from the earlier admission.’” 564 F.3d at 1175-76
(citations omitted). Nevertheless, the Ninth Circuit has made clear that “[i]t makes no
difference that . . . the misquoted language is supported generally by [the cited case].”
Malkeet Lnu v. Blanche, 177 F.4th 1014, 1027 (9th Cir. 2026).
Case 2:23-cv-02090-DWL Document 172 Filed 07/28/26 Page 5 of 6

- 6 -
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Accordingly,
IT IS ORDERED that within 14 days from the date of this order, Ms. Tate shall
file a supplemental memorandum, not to exceed 10 pages, addressing the issues raised in
this order.
Dated this 28th day of July, 2026.
Case 2:23-cv-02090-DWL Document 172 Filed 07/28/26 Page 6 of 6