Mata v. Avianca, Inc.

Filing 45

MEMORANDUM OF LAW re: 33 Order to Show Cause,,,,,,,,,,, . Document filed by Levidow, Levidow & Oberman, P.C., Steven Schwartz..(Maulsby, Tyler)

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 1 of 29

UNITED STATES DISTRICT COURT
SOUTHERN DISTRICT OF NEW YORK
--------------------------------------------------------------X
ROBERTO MATA,
:
:
Plaintiff,
: Case No.: 22-cv-1461 (PKC)
:
- against –
:
:
AVIANCA, INC.,
:
:
Defendant.
:
--------------------------------------------------------------X

MEMORANDUM OF LAW BY NON-PARTIES STEPHEN A. SCHWARTZ AND
LEVIDOW LEVIDOW & OBERMAN, P.C. IN
RESPONSE TO MAY 26, 2023 ORDER TO SHOW CAUSE

FRANKFURT KURNIT KLEIN & SELZ, P.C.
Tyler Maulsby
Ronald C. Minkoff
Ashley K. Alger
28 Liberty Street, 35th Floor
New York, New York 10005
Tel.: [phone]
Fax: [phone]
[e-mail]
[e-mail]
[e-mail]
Attorneys for Non-Parties Stephen A. Schwartz,
Esq. and Levidow Levidow & Oberman, P.C.

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 2 of 29

TABLE OF CONTENTS
Page
PRELIMINARY STATEMENT .................................................................................................... 1
RELEVANT FACTUAL BACKGROUND ................................................................................... 3
ARGUMENT ................................................................................................................................ 11
I.

II.

THE RELEVANT LAW APPLICABLE TO SANCTIONS............................................ 11
A.

Sua Sponte Sanctions Under Rule 11.................................................................... 11

B.

28 U.S.C. § 1927 And The Court’s Inherent Authority ........................................ 13

SANCTIONS ARE NOT WARRANTED BECAUSE RESPONDENTS
ACTIONS WERE NOT IN BAD FAITH ........................................................................ 14
A.

Mr. Schwartz’s Misunderstanding of ChatGPT Does Not Show Bad Faith ........ 15

B.

Mr. Schwartz Promptly Acknowledged His Mistake ........................................... 17

C.

The Firm Should Not Be Sanctioned .................................................................... 18

D.

Mr. LoDuca’s Actions Do Not Support a Finding of Sanctions ........................... 18

E.

Mr. Schwartz, Mr. LoDuca, and the Firm Acted Reasonably .............................. 20

III.

THE NOTARIZATION OF THE APRIL 25 AFFIDAVIT WAS NOT FALSE
OR FRAUDULENT ......................................................................................................... 22

IV.

SANCTIONS ARE UNNECESSARY ............................................................................. 23

CONCLUSION ............................................................................................................................. 24

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 3 of 29

TABLE OF AUTHORITIES
Page(s)
Cases
Braun ex rel. Advanced Battery Techs., Inc. v. Zhiguo Fu,
No. 11-CV-04383, 2015 WL 4389893 (S.D.N.Y. July 10, 2015) ...............................13, 14, 19
In re Bees,
562 F.3d 284 (4th Cir. 2009) ...................................................................................................22
Chambers v. NASCO, Inc.,
501 U.S. 32 (1991) .............................................................................................................13, 14
De La Fuente v. DCI Telecommunications, Inc.,
269 F. Supp. 2d 229 (S.D.N.Y.), aff’d and remanded, 82 F. App’x 723 (2d
Cir. 2003) .................................................................................................................................20
Fishoff v. Coty Inc.,
634 F.3d 647 (2d Cir. 2011).....................................................................................................19
Gallop v. Cheney,
667 F.3d 226 (2d Cir. 2012) (amended on Feb. 3, 2012) ........................................................20
Huebner v. Midland Credit Mgmt., Inc.,
897 F.3d 42 (2d Cir. 2018).......................................................................................................13
Miller v. United Airlines, Inc.,
174 F.3d 366 (2d Cir. 1999).......................................................................................................6
Muhammad v. Walmart Stores East, L.P.,
732 F.3d 104 (2d Cir. 2013).....................................................................................................11
New V & J Produce Corp. v. NYCCaterers Inc.,
No. 13 CIV. 4861 ER, 2014 WL 5026157 (S.D.N.Y. Sept. 29, 2014) ....................................17
Olive Grp. N. Am. LLC v. Afghanistan Int'l Bank,
No. 21-CV-10836 (ER), 2023 WL 2644350 (S.D.N.Y. Mar. 27, 2023) .................................13
In re Pennie & Edmonds LLP,
323 F.3d 86 (2d. Cir. 2003)..........................................................................................11, 12, 20
Rankin v. City of Niagara Falls,
293 F.R.D. 375 (W.D.N.Y. 2013), aff'd, 569 F. App'x 25 (2d Cir. 2014)
................................................................................................................................12, 17, 18, 23
Rivas v. Bowling Green Assocs., L.P.,
No. 13-CV-7812 PKC, 2014 WL 3694983 (S.D.N.Y. July 24, 2014) .......................12, 15, 24
ii

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 4 of 29

United States v. Int'l Bhd. of Teamsters, Chauffeurs, Warehousemen & Helpers of
Am., AFL-CIO,
948 F.2d 1338 (2d Cir. 1991).............................................................................................13, 17
Varghese v. China Southern Airlines Co., Ltd.,
925 F.3d 1339 (11th Cir. 2009) ...................................................................................6, 7, 8, 10
Weddington v. Sentry Indus., Inc.,
No. 18-CV-10055 (PKC), 2020 WL 264431 (S.D.N.Y. Jan. 17, 2020) ..................................15
Zicherman v. Korean Air Lines Co., Ltd.,
516 F.3d 1237 (11th Cir. 2008) .............................................................................................6, 7
Statutes
28 U.S.C. 1441(a) ............................................................................................................................4
28 U.S.C. § 1441(b)(2) ..................................................................................................................15
28 U.S.C. § 1927 ............................................................................................................1, 13, 18, 24
Other Authorities
Fed. R. Civ. P. 11 ................................................................................................................... passim
James Vincent, OpenAI Isn't Doing Enough to Make ChatGPT's Limitations
Clear, The Verge (May 30, 2023, 11:22 AM EDT),
https://www.theverge.com/2023/5/30/23741996/openai-chatgpt-falseinformation-misinformation-responsibility................................................................................7
Judge Herbert B. Dixon Jr. (Ret.), My “Hallucinating” Experience with
ChatGPT, Judges' J., Spring 2023, at 37,
https://www.americanbar.org/content/dam/aba/publications/judges_journal/vo
l62no2-jj2023-tech.pdf.............................................................................................................21
N.Y. Rule of Prof'l Conduct 1.1, Comm. 8 ....................................................................................21

iii

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 5 of 29

Non-Parties Stephen A. Schwartz, Esq. (“Mr. Schwartz”) and Levidow Levidow &
Oberman, P.C. (the “Firm”), by their attorneys Frankfurt Kurnit Klein & Selz P.C., submit this
Memorandum of Law in response to the Court’s Order to Show Cause dated May 26, 2023
(Docket Entry (“DE”) 33, the “Order”), which directs Mr. Schwartz and the Firm to address
whether the Court should sanction them pursuant to Fed. R. Civ. P. 11(b)(2) and (c) (“Rule 11”),
28 U.S.C. § 1927 (“Section 1927”), and the Court’s inherent authority.
PRELIMINARY STATEMENT
In the Order, the Court describes this situation as “unprecedented.” We agree. We can
find no case where, as here, a lawyer using a new, highly-touted research tool obtained cases that
the research tool itself completely made up. The lawyer, Mr. Schwartz, had no idea this was
happening, even when opposing counsel brought their inability to locate the cases to his
attention. ChatGPT even assured him the cases were real and could be found on Westlaw and
LexisNexis, and continued to provide extended excerpts and favorable quotations. Now that Mr.
Schwartz and the Firm know ChatGPT was simply making up cases, they are truly mortified;
they had no intention of defrauding the Court, and the mere accusation – repeated in hundreds (if
not thousands) of articles and online posts – has irreparably damaged their reputations. They
have apologized to the Court in earlier submissions and do so again here.
Sanctions, however, are another matter altogether. The Second Circuit case law is clear:
whether under Rule 11, Section 1927 or the Court’s inherent authority, Mr. Schwartz and the
Firm may be sanctioned only if they acted with subjective bad faith, that is, if they actually knew
the case law was false and provided it in order to defraud the Court. That did not happen here.
Mr. Schwartz, a personal injury and workers compensation lawyer who does not often practice in
federal court, found himself researching a bankruptcy issue under the Montreal Convention 1999
(the “Montreal Convention”). He also found that his firm’s Fastcase subscription no longer

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 6 of 29

worked for federal searches. With no Westlaw or LexisNexis subscription, he turned to
ChatGPT, which he understood to be a highly-touted research tool that utilizes artificial
intelligence (AI). He did not understand it was not a search engine, but a generative language
processing tool primarily designed to generate human-like text response based on the user’s text
input and the patterns it recognized in data and information used during its development or
“training,” with little regard for whether those responses are factual. Given that the technology
is so new, the press coverage so favorable, and the warnings on ChatGPT’s website so vague
(particularly at the time Mr. Schwartz used it), his ignorance was understandable. While,
especially in the light of hindsight, he should have been more careful and checked ChatGPT’s
results, he certainly did not intend to defraud the Court – a fraud that any lawyer would have
known would be quickly uncovered, as this was. There was no subjective bad faith here.
Nor was there any misconduct related to the notarization of the April 25, 2023 Affidavit.
Mr. Schwartz and his associate Peter LoDuca (“Mr. LoDuca”) have sworn under oath that the
Affidavit was created on April 25; that Mr. LoDuca signed it in Mr. Schwartz’s presence; and
that Mr. Schwartz signed it and affixed his notary stamp to it that same day. The fact that the
notary portion of the affidavit says it was dated “January” rather than “April” resulted from a
clerical error. This was not sanctionable misconduct.
Finally, sanctions would serve no useful purpose. Mr. Schwartz and the Firm have
already become the poster children for the perils of dabbling with new technology; their lesson
has been learned. The Firm has taken and is taking a series of remedial steps: obtaining better
research tools for its lawyers; implementing firm-wide CLEs in technology; imposing policies
against using AI tools without checking, and more. At this point, any additional sanctions the
Court imposes will be merely, and unnecessarily, punitive.

2

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 7 of 29

No sanctions are warranted.
RELEVANT FACTUAL BACKGROUND
Respondents’ Background
Steven A. Schwartz is a member of the Firm and has been practicing law in New York
since 1992. Declaration of Steven A. Schwartz dated June 6, 2023 (“Schwartz Decl.”) ¶ 5. Mr.
Schwartz graduated from the State University of New York (SUNY) at Albany in 1988 and
received his Juris Doctor degree from New York Law School in 1991. Id. Mr. Schwartz has
worked his entire career for the Firm, where he practices in the areas of workers compensation
and personal injury on behalf of injured claimants under the New York Workers Compensation
Law. Id. at ¶¶ 5-6. Because his practice is mainly limited to state court and administrative
bodies, he is not a member of the Bar of this Court and rarely appears here. Id. at ¶ 6.
Peter LoDuca received his law degree from St. John’s University in 1985 and has been
employed as an associate of the Firm since 1996. Declaration of Peter LoDuca dated June 6,
2023 (“LoDuca Decl.”) ¶ 2. Neither Mr. Schwartz nor Mr. LoDuca have ever been disciplined
by any court or tribunal. Id.; Schwartz Decl. ¶ 6.
Levidow, Levidow & Oberman, P.C. is a small firm based in New York City that focuses
primarily on workers compensation and personal injury matters in New York State courts as well
as before the New York State Workers Compensation Board. Declaration of Thomas Corvino
dated June 6, 2023 (“Corvino Decl.”) ¶ 4. The Firm and its lawyers do not regularly litigate in
federal court. Id. at ¶ 4, 6. The Firm has never been the subject of discipline or sanctions in any
proceeding. Id. at ¶ 4.
The Instant Action and Defendant’s Motion to Dismiss
On July 28, 2020, Plaintiff, Roberto Mata, filed a summons and complaint through
counsel in the Supreme Court of New York, New York County, alleging that he suffered
3

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 8 of 29

personal injuries while aboard a flight operated by Defendant, Avianca, Inc. (DE 21-1.) Mr.
Mata’s primary counsel is Mr. Schwartz, who filed the state court summons and complaint. The
state court action was automatically stayed when Avianca filed for bankruptcy. (DE 21, ¶ 8.)
On January 31, 2022, after Avianca’s bankruptcy had concluded and the automatic stay lifted,
the parties agreed to voluntarily discontinue the state court action so that it could be re-filed.
Two days later, on February 2, 2022, Mr. Mata, with Mr. Schwartz still his counsel of record, refiled his action against Avianca in state court. On February 22, 2022, Avianca removed this
action pursuant to, inter alia, 28 U.S.C. 1441(a). (DE 1.)
Upon removal, Mr. Schwartz, who is not a member of this Court, could no longer appear
as counsel of record. Mr. LoDuca, who is a member of this Court, appeared instead, while Mr.
Schwartz continued to work on the matter and handle the substantive legal research and writing.
(See DE 32 at ¶ 3.) See Schwartz Decl. ¶ 8. Avianca moved to dismiss the claims as time barred
by the two year limitations period set by the Montreal Convention. (DE 16.)
Respondents’ Opposition to the Motion to Dismiss
Respondents submitted an Affirmation in Opposition to Avianca’s motion to dismiss, in
which they argued that Mr. Mata’s claims were timely filed because either: (i) the Montreal
Convention did not preempt New York’s three-year statute of limitations; or (ii) even if the
Montreal Convention’s two-year statute of limitations applied, the claims were still timely filed.
(DE 21.)
The Affirmation in Opposition to the motion to dismiss was prepared by Mr. Schwartz,
but was signed and filed by Mr. LoDuca. Schwartz Decl. ¶ 8. In drafting the opposition papers,
Mr. Schwartz attempted to perform legal research on the interplay between U.S. bankruptcy law
and the Montreal Convention, an issue with which he was not familiar. Schwartz Decl. ¶ 9.

4

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 9 of 29

The Firm’s primary research database is Fastcase, an online legal research subscription
that is available to all lawyers at the Firm. Corvino Decl. ¶ 8. The Firm does not maintain a
subscription to Westlaw or LexisNexis, which are much more costly. Id. Prior to this incident,
the Firm was under the impression that it maintained a subscription in Fastcase that allowed its
lawyers to search both state and federal cases in New York. Corvino Decl. ¶¶ 8-10. However,
when Mr. Schwartz attempted to conduct research on Fastcase for this matter, he was unable to
access the federal database. Schwartz Decl. ¶ 12. The Firm later learned that because of a
billing error, the Firm’s Fastcase access to the federal case database had been inadvertently
deactivated. Corvino Decl. ¶ 10.1 The Firm has since rectified this issue and has made clear to
all its lawyers that they are able to access both databases. Id. at ¶¶ 10-11.
Mr. Schwartz Uses ChatGPT To Assist With Legal Research
Because Mr. Schwartz’s Fastcase access was limited, he turned to ChatGPT to conduct
additional legal research. Schwartz Decl. ¶ 13. As detailed below, Mr. Schwartz did not fully
understand ChatGPT. Mr. Schwartz had never before used ChatGPT for legal research; he knew
of the technology from his college-aged children. Id. at ¶ 14. Mr. Schwartz also remembered
reading several articles touting the benefits of artificial intelligence tools and their use in
professional settings, including at law firms. Id. Indeed, as discussed below, during the early
part of 2023 when Mr. Schwartz conducted his research, such articles were ubiquitous in the
legal and public press, with some predicting that AI tools such as ChatGPT could make legal
research obsolete.2
1

To be clear, the Firm was able to access some federal cases through Fastcase, which is why it was able to provide
two of the requested cases. (See DE 29-6 and 29-8.) However, the limitation made it so that Mr. Schwartz and the
Firm’s lawyers could not run full-fledged searches for federal cases.
2

A small but representative sample of these articles is attached as Exhibit A to the Declaration of Ashley K. Alger
(“Alger Decl.”). ¶ 2.

5

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 10 of 29

Mr. Schwartz therefore decided to use ChatGPT to find additional support for his
arguments under the Montreal Convention. Mr. Schwartz approached this task the way a lawyer
would generally approach a legal research project using a standard database – unaware that
ChatGPT involved a very different technology. He first asked a broad question about tolling
under the Montreal Convention, then asked a more targeted question and requested case law.
Schwartz Decl. Ex. A at 1-2.3 ChatGPT answered that “[t]here are several U.S. court cases that
have held that the statute of limitations is tolled by the bankruptcy of a defendant pursuant to the
Montreal Convention. Here are a few examples. . . .” Id. at 2. ChatGPT listed three cases as
“examples” for this proposition, complete with case citations, summaries and lengthy, very
favorable quotations. Id. Each time Mr. Schwartz asked for specific case support, ChatGPT
answered “Certainly!” and would provide Mr. Schwartz with what it said was a “brief excerpt”
from the opinion. Id. at 8, 10, 12-14. For example, in response to Mr. Schwartz’s request to
“show [him]” the Varghese case, ChatGPT responds: “Certainly! Here’s a brief excerpt from
the opinion of the United States Court of Appeals for the Eleventh Circuit in Varghese v. China
Southern Airlines Co., Ltd., 925 F.3d 1339 (11th Cir. 2009)”). Id. at 12. See also id. at 13
(ChatGPT providing brief excerpt from Zicherman v. Korean Air Lines Co., Ltd., 516 F.3d 1237
(11th Cir. 2008)); id. at 14 (ChatGPT providing Miller v. United Airlines, Inc., 174 F.3d 366 (2d
Cir. 1999)).
As noted, Mr. Schwartz did not fully understand how generative pre-trained transformer
(GPT) technology worked. Mr. Schwartz believed that ChatGPT was a search engine – similar
to Google or even a search database such as Westlaw or LexisNexis – that provided answers to

3

A full history of Mr. Schwartz’s ChatGPT conversation is included as Exhibit A to his Declaration. Schwartz
Decl. ¶ 18.

6

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 11 of 29

questions using natural language based on publicly available information, including publicly
available case law. Schwartz Decl. ¶ 15, 20. Mr. Schwartz recognizes now that this is incorrect.
Id. at ¶ 29. He had no idea at the time, however, that in response to a question ChatGPT could
generate an answer that, while intended to converse with and inform its user, was complete
fiction – and a cleverly disguised fiction to boot.4 Id. at ¶ 30. Had he known that, he certainly
would not have used the technology for these purposes. Id. at ¶ 32, 36.
The April 25 Affidavit
In its reply brief, Defendant’s counsel stated that it could not locate several cases cited in
Respondents’ opposition.5 (DE 24.) On April 11 and 12, 2023, this Court ordered Mr. LoDuca
to file an affidavit annexing eight cases that defense counsel could not locate. (DE 25, 27.)
Mr. Schwartz and Mr. LoDuca prepared an affidavit in response to the Court’s April 11
and April 12 Orders, which Mr. LoDuca filed on April 25, 2023. (DE 29, the “April 25
Affidavit.”) The April 25 Affidavit purported to annex copies of the eight cases listed in the
Court’s Orders dated April 11, 2023 and April 12, 2023.6 (DE 25, 27, 29.) Mr. Schwartz printed
two of the eight cases from Fastcase (see DE 29-6 and 29-8) and returned to ChatGPT for the
4

See James Vincent, OpenAI Isn’t Doing Enough to Make ChatGPT’s Limitations Clear, The Verge (May 30, 2023,
11:22 AM EDT), https://www.theverge.com/2023/5/30/23741996/openai-chatgpt-false-information-misinformationresponsibility (noting that the media often fails to convey the inaccurate nature of such platforms and that
ChatGPT’s disclosures are inadequate to warn users as well). Nor did the warning on ChatGPT’s homepage – that
ChatGPT “may occasionally generate incorrect information” – alert Mr. Schwartz to the dangers. See Alger Decl.,
Ex. B. That statement could apply to virtually any web searches. It hardly would lead a reasonable person to
believe the technology would make cases up out of whole cloth. We understand that the first time someone uses
ChatGPT, additional warnings appear. However, given that Mr. Schwartz had once or twice before used ChatGPT
for leisure (when his children were showing him the program), it is likely those same warnings did not appear. Mr.
Schwartz has no recollection of seeing those warnings before using ChatGPT for research.
5

Six of the ten cases were cited in Respondent’s Opposition. (DE 24.) The other four cases were quoted or
mentioned in one of the cited cases, Varghese v. China Southern Airlines Co., Ltd., 925 F.3d 1339 (11th Cir. 2019).
(Id.)
6

The April 25 Affidavit stated that Respondents “could not locate” Zicherman v. Korean Air Lines Co., Ltd., 516
F.3d 1237 (11th Cir. 2008), which was one of the requested cases and was cited within one of the other cases
(Varghese) that turned out to be fabricated.

7

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 12 of 29

remaining six cases (the “ChatGPT Cases”). Schwartz Decl. ¶ 24. Again, at this time Mr.
Schwartz was under the misimpression that ChatGPT was a reliable search engine, not an
experimental project in generative AI. Id.; see also id. at ¶¶ 15, 20, 29. Mr. Schwartz therefore
asked ChatGPT to provide copies of the six cases, which Mr. Schwartz and Mr. LoDuca annexed
to the April 25 Affidavit what ChatGPT provided. Schwartz Decl. ¶ 25. The April 25 Affidavit
acknowledged that the opinions for the six ChatGPT Cases “may not be inclusive of the entire
opinions but only what is made available by online database.” (DE 29.) As detailed above, Mr.
Schwartz did not have immediate access to a more comprehensive research tool such as Westlaw
or LexisNexis (or even a complete Fastcase subscription). Nevertheless, he reasonably believed
that ChatGPT was providing results based on publicly available information. Schwartz Decl. at ¶
12-13, 15, 20. He certainly had no intention of misleading his adversaries or the Court. Id. at ¶
36.
Indeed, the ChatGPT Cases were not immediately suspect. For example, for the case
Varghese v. China Southern Airlines Co. Ltd., the attached document included: a case caption
indicating that the decision came from the United States Court of Appeals for the Eleventh
Circuit; a docket number that, at first glance, tracks the format of circuit court appellate case
numbers; and a listing of three circuit judges.7 (See DE 29-1.) Another document purporting to
be Petersen v. Iran Air indicated that the decision was issued by the United States District Court
for the District of Columbia (the court was listed at the top of the page), listed the issuing judge
(a current District Judge in the District of Columbia), and contained both a federal reporter

7

Of the three circuit judges listed – Jordan, Rosenbaum, and Higginbotham – the first two sit on the Eleventh
Circuit (and were active on the date listed as the decision date for the Varghese opinion). See Eleventh Circuit
Judges, United States Court of Appeals for the Eleventh Circuit, https://www.ca11.uscourts.gov/eleventh-circuitjudges. The third judge, Higginbotham, is a Fifth Circuit judge, however, the opinion at DE 26-1 shows an asterisk
next to Judge Higginbotham’s name, which could reasonably suggest that he was sitting by designation.

8

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 13 of 29

citation and a Lexis citation. (See DE 29-3.) The remaining attachments to the April 25
Affidavit contained similar indicia of authenticity, including case captions, the identities of the
judges and lawyers involved in the case, and internal citations throughout the opinions. (See DE
29-2, 4, 5, 7.)
Mr. Schwartz and the Firm now recognize that the Court’s and their adversary’s inquiries
should have been a red flag, and that they should have taken more affirmative steps at that point
– e.g., by contacting a law library or colleague with Westlaw or LexisNexis access – to check
ChatGPT’s results, instead of going back to ChatGPT itself. Schwartz Decl. ¶ 25, 33; Corvino
Decl. ¶ 12. Yet their explanation is simple and reasonable: it still seemed inconceivable to them
that ChatGPT’s results were complete fabrications.
The First Order to Show Cause and Respondents’ Reply
On April 26, 2023, Defendant’s counsel filed a letter noting that it still could not locate
many of the cases. (DE 30.) On May 4, 2023, the Court issued an Order (DE 31, the “First
OSC”), in which the Court observed that six of the proffered cases in the April 25 Affidavit (i.e.,
the ChatGPT Cases) appeared to be fabricated. The First OSC directed Mr. LoDuca to show
cause why he should not be sanctioned for providing six decisions which did not exist.
When he read the First OSC, Mr. Schwartz finally realized there had to be a serious
problem with the tool he had used to conduct legal research.8 Schwartz Decl. ¶ 29. Mr.
Schwartz immediately conducted further research into ChatGPT and learned that it was
inherently unreliable for anything related to legal research and analysis. Id. at ¶ 30. As part of
this process, Mr. Schwartz submitted additional queries to ChatGPT. For example, he asked

8

Mr. LoDuca did not personally conduct any of the legal research involved in the Affirmation in Opposition to the
motion to dismiss nor did he have personal knowledge of how the legal research was conducted by Mr. Schwartz.
(See DE 32 at ¶ 4.)

9

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 14 of 29

ChatGPT whether Varghese, which at that point Mr. Schwartz understood did not exist (see DE
31), was a real case. ChatGPT confidently answered “Yes, Varghese . . . is a real case.” (DE 321). Mr. Schwartz asked ChatGPT to verify its source for the Varghese case. ChatGPT doubleddown and stated that the case “does indeed exist and can be found on legal research databases
such as Westlaw and LexisNexis.” (Id.) Mr. Schwartz did not submit these questions to
ChatGPT for the purpose of trying to convince the Court that the ChatGPT Cases existed.
Schwartz Decl. ¶ 30. Rather, knowing he had made a mistake, he wanted to show the lengths to
which ChatGPT would go to mislead him, something he never expected based on his
understanding of the technology. Id. at ¶ 31.
In response to the First OSC, Mr. LoDuca submitted an affidavit (DE 32, the “LoDuca
Affidavit”), which also included an affidavit from Mr. Schwartz (the “Schwartz Affidavit”) as an
attachment. (See DE 32-1.) Mr. LoDuca stated that Mr. Schwartz was primarily responsible for
conducting the research and drafting the opposition to the original motion to dismiss and was
responsible for finding the ChatGPT Cases. (DE 32 at ¶¶ 3-4.) Mr. LoDuca further attested that
owing to his long professional relationship with Mr. Schwartz, as well as Mr. Schwartz’s
seniority and position at the Firm, Mr. LoDuca had no reason to doubt the authenticity of the
cases that Mr. Schwartz provided. (Id. at ¶ 6.)
The Schwartz Affidavit confirmed that Mr. Schwartz was solely responsible for drafting
the opposition to the motion to dismiss and conducting the legal research. (DE 32-1 at ¶ 5.) Mr.
Schwartz further confirmed that he had located the six non-existent cases using ChatGPT, which
he did not know was unreliable. (Id. at ¶¶ 6-11.) Mr. Schwartz also acknowledged that it was his
fault for “not confirming the sources provided by ChatGPT” but that he never had any intention
to mislead the Court. (Id. at ¶¶ 11, 10.) Mr. Schwartz further attested that he “greatly regrets

10

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 15 of 29

having utilized generative artificial intelligence to supplement the legal research performed
herein and will never do so in the future without absolute verification of its authenticity.” (Id. at
¶ 13.)
On May 26, 2023, the Court issued the Order directing Mr. Schwartz and the Firm to
show cause why they should not be sanctioned for providing the ChatGPT Cases. (DE 33.)
ARGUMENT
I.

THE RELEVANT LAW APPLICABLE TO SANCTIONS
In order to impose sanctions here, the Court must find that Respondents acted with

subjective bad faith. The standard is the same whether under Rule 11, Section 1927 or the
Court’s inherent authority. As detailed below, Respondents’ conduct does not rise to the level of
subjective bad faith.
A.

Sua Sponte Sanctions Under Rule 11

Rule 11(b)(2) provides that where an attorney presents a paper to the Court, the attorney
certifies to the best of his knowledge, information and belief that “the claims, defenses, and other
legal contentions are warranted by existing law or by a nonfrivolous argument for extending,
modifying, or reversing existing law or for establishing new law.” Rule 11(c)(3) further provides
that in lieu of a motion for sanctions filed by a party, the Court on its own motion “may order an
attorney, law firm, or party to show cause why conduct specifically described in the order has not
violated Rule 11(b).”
The standard for a court to impose sanctions sua sponte under Rule 11(c)(3) is
significantly higher than the standard under Rule 11(c)(2) in response to a motion filed by a
party. The law in this Circuit is clear: imposing sua sponte sanctions under Rule 11 requires a
finding of subjective bad faith, not just that the conduct was objectively unreasonable. See In re
Pennie & Edmonds LLP, 323 F.3d 86, 90 (2d. Cir. 2003); Muhammad v. Walmart Stores East,
11

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 16 of 29

L.P., 732 F.3d 104, 108 (2d Cir. 2013); see also; Rivas v. Bowling Green Assocs., L.P., No. 13CV-7812 PKC, 2014 WL 3694983, at *1 (S.D.N.Y. July 24, 2014) (Castel, J.) (“in the case of
court-initiated Rule 11 sanctions . . . the standard to be applied is not one of objective
unreasonableness, but subjective bad faith”). A finding of subjective bad faith under Rule
11(c)(3) “must meet the high evidentiary burden of clear and convincing evidence applicable to
findings of contempt . . ..” Rankin v. City of Niagara Falls, 293 F.R.D. 375, 387-88 (W.D.N.Y.
2013), aff'd, 569 F. App'x 25 (2d Cir. 2014) (citations omitted).
The policy rationale behind the heightened standard for sua sponte sanctions makes
sense. Unlike with a motion under Rule 11(c)(2), Rule 11(c)(3) does not afford the respondent
the benefit of the 21-day safe harbor period to withdraw the allegedly frivolous pleading. See In
re Pennie, 323 F.3d at 90 (“because the ‘safe harbor’ protection does not exist when a lawyer’s
submission is challenged in a show cause proceeding initiated by a trial judge, the more rigorous
standard of bad faith should apply.”). As detailed above, as soon as the Court issued the First
OSC, Respondents recognized their error and were forthcoming with the Court. (See DE 32, 321.) Although Respondents arguably should have realized their error sooner, this was not a
scenario where they had the benefit of the safe harbor period as contemplated in Rule 11.
Accordingly, the subjective bad faith standard reserved for Rule 11(c)(3) is appropriate here.
As this Court has previously held, subjective bad faith in the context of sanctions requires
the Court to find that the respondents acted with actual knowledge that the conduct was
frivolous. See Rivas, 2014 WL 3694983, at *2 (“[p]roof of actual knowledge, and not merely
what a reasonable attorney should have known, is required.”); see also In re Pennie, 323 F.3d at
88 (record did not establish subjective bad faith despite findings by District Court that attorneys
allowed client to submit a partially fraudulent affidavit and that “the law firm could not have

12

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 17 of 29

objectively believed [the statements] were true”); Braun ex rel. Advanced Battery Techs., Inc. v.
Zhiguo Fu, No. 11-CV-04383, 2015 WL 4389893, at *14 (S.D.N.Y. July 10, 2015) (“subjective
bad faith standard, as applied in the context of sua sponte Rule 11 sanctions, requires an attorney
to have actual knowledge that a pleading or argument that he or she is advancing is frivolous.”).
B.

28 U.S.C. § 1927 And The Court’s Inherent Authority

Section 1927 provides courts the power to sanction attorney conduct that is
“unreasonable and vexatious” and leads to the “multiplications of proceedings.” Olive Grp. N.
Am. LLC v. Afghanistan Int'l Bank, No. 21-CV-10836 (ER), 2023 WL 2644350 at *5 (S.D.N.Y.
Mar. 27, 2023). Sanctions under Section 1927 are intended to “deter unnecessary delays in
litigation” but are only proper when the attorney’s actions are “so completely without merit as to
require the conclusion that they must have been undertaken for some improper purpose[.]”
United States v. Int'l Bhd. of Teamsters, Chauffeurs, Warehousemen & Helpers of Am., AFLCIO, 948 F.2d 1338, 1345 (2d Cir. 1991). Implicit in Section 1927 is the requirement that a party
engage in some intentional wrongdoing that wastes time and resources of the court and the
parties. See Huebner v. Midland Credit Mgmt., Inc., 897 F.3d 42, 55 (2d Cir. 2018) (Section
1927 requires a court to find clear evidence that “(1) the offending party’s claims were entirely
without color” and “(2) the claims were . . . motivated by improper purposes such as harassment
or delay.”). To impose sanctions under Section 1927, the Court must find that the attorney acted
in bad faith. See Teamsters, 948 F.2d at 1345, 47 (“Both § 1927 and the court’s inherent power
demand, inter alia, a clear demonstration of bad faith in order to justify sanctions.”).
The court’s inherent authority to issue sanctions, which stems from the court’s need to
“manage [its] own affairs” and effectuate the “orderly and expeditious disposition of cases” is
likewise only appropriate when there is a “particularized showing of bad faith.” Chambers v.
NASCO, Inc., 501 U.S. 32, 43 (1991); Teamsters, 948 F.2d at 1345. The Supreme Court has
13

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 18 of 29

cautioned that the court’s inherent authority should be exercised “with restraint and discretion”
and that sanctions are proper only when the party being sanctioned has knowingly acted in bad
faith. Chambers, 501 U.S. at 44; see also Braun, 2015 WL 4389893, at *17 (inherent authority
sanctions are inappropriate unless party “knowingly submitted a materially false or misleading
pleading, or knowingly failed to correct false statements, as part of a deliberate and
unconscionable scheme to interfere with the Court’s ability to adjudicate the case fairly.”).
Sanctions imposed under the court’s inherent authority require a finding of bad faith “by clear
and convincing evidence” and the court’s findings of fact “must be characterized by a high
degree of specificity.” Braun, 2015 WL 4389893, at *17.
II.

SANCTIONS ARE NOT WARRANTED BECAUSE RESPONDENTS ACTIONS
WERE NOT IN BAD FAITH
This Court should decline to impose sanctions because Respondents did not exhibit

subjective bad faith, which is required under Rule 11, Section 1927 and inherent authority. The
absence of bad faith is evident for at least three reasons. First, Mr. Schwartz’s actions were
based on a fundamental misunderstanding of what ChatGPT is and how it worked. As detailed
below, Mr. Schwartz legitimately thought that he was using a new, state-of-the art search engine,
not an artificial language program. He had no idea this program would fabricate case law, and
certainly was not trying to mislead the court or knowingly engage in frivolous or vexatious
conduct. Second, Mr. Schwartz could not reasonably have been expected to know that ChatGPT
would make up entire cases and then continue to lie to him even after he questioned the its
results. Third, Mr. Schwartz promptly acknowledged his mistake as soon as he realized what had
transpired. These actions are inconsistent with evidence of bad faith behavior.

14

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 19 of 29

A.

Mr. Schwartz’s Misunderstanding of ChatGPT Does Not Show Bad Faith

For sanctions to issue, the Court must find that Mr. Schwartz knew that he was submitting
fictitious cases to the Court, either in the original Affirmation in Opposition or the April 25
Affidavit. Rivas, 2014 WL 3694983, at *2 (“Proof of actual knowledge, and not merely what a
reasonable attorney should have known, is required” to satisfy the subjective bad faith standard)
(emphasis added). As detailed above, Mr. Schwartz did not know that the ChatGPT Cases were
fictitious. On the contrary, because he did not understand how ChatGPT worked, Mr. Schwartz
believed that ChatGPT had gathered legitimate cases from publicly available sources. He did not
think for a second that he was using a technology that was capable of fabricating information or
lying to him about the authenticity of the information it provided.
This Court’s decision in Rivas is instructive here. In Rivas, this Court ordered attorneys
to show cause why they should not be sanctioned pursuant to Rule 11(b) for filing a notice of
removal that was clearly prohibited by 28 U.S.C. § 1441(b)(2). 2014 WL 3694983, at *1. The
attorneys in Rivas were well aware that 28 U.S.C. § 1441(b)(2) foreclosed their client’s ability to
remove the case but filed a notice of removal anyway in the hopes that the plaintiff would not
challenge it. Id. at *2. In other words, the attorneys in Rivas knew that their actions were
prohibited (i.e. filing a notice of removal), but did so anyway. As a result, the Court found that
the attorneys’ intentional conduct evidenced subjective bad faith. Id. at *4.9
Here, Mr. Schwartz’s conduct could not be more different. As detailed above, Mr.
Schwartz did not know that the ChatGPT cases were fabricated when he submitted them to the

9

The same is true of this Court’s decision in Weddington v. Sentry Indus., Inc., No. 18-CV-10055 (PKC), 2020 WL
264431 (S.D.N.Y. Jan. 17, 2020), in which this Court sanctioned attorneys for moving to dismiss based on the
residency of their client despite knowing that their statements in the motion about their own client’s domicile were
false. Unlike here, the attorneys in Weddington had multiple opportunities to fix their deficient pleading but
declined to do so. The error in Weddington was also far more fundamental than what happened here; Mr. Schwartz
was using a new technology that he did not understand; he was not choosing to ignore obvious adverse facts.

15

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 20 of 29

Court. Mr. Schwartz’s understanding of ChatGPT at the time was that it was a search engine,
not a “large language model” designed to simulate human conversation regardless of the
accuracy of the information. This is a common misunderstanding. The articles attached as
Exhibit C to the Alger Declaration make clear that one of the biggest risks of this type of
technology is that a significant segment of the population (including other members of the legal
profession) does not fully understand how it works or where it can go wrong. At the time Mr.
Schwartz used ChatGPT to conduct legal research, the technology was the subject of almost
daily discussion in the media and the legal industry. See Alger Decl., Ex. A (articles from
January and February 2023). Indeed, many articles circulated within the legal industry at the
time focused on whether ChatGPT would revolutionize how lawyers and law firms operate, and
touted ChatGPT as a legitimate tool for lawyers to use in certain circumstances. See id.
Moreover, the history of Mr. Schwartz’s interactions with ChatGPT in this case weighs
against a finding of subjective bad faith. As detailed above, Mr. Schwartz did not simply enter a
single query into ChatGPT and paste the answer into his opposition. Instead, he used the
technology (albeit incorrectly) as though it was a legitimate research tool. Mr. Schwartz asked
for information about general legal issues, cases that stood for a certain proposition, and even
copies of the cases that ChatGPT was summarizing and citing to in the first instance. See
Schwartz Decl. Ex. A. Each time, ChatGPT provided answers that appeared legitimate, or at
least not obviously false. Indeed, even when Mr. Schwartz directly asked ChatGPT whether the
cases it had provided were fake, the program not only said the cases were real, but also assured
Mr. Schwartz that the cases were available on Westlaw and LexisNexis. Id. At the time, Mr.
Schwartz did not think he needed to doubt this assurance.

16

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 21 of 29

Of course, Mr. Schwartz realizes in hindsight that he missed several warning signs and
should have taken other steps to verify the accuracy of the cases he was citing. For example, Mr.
Schwartz recognizes that he should not have tried to use ChatGPT in the first place – a new
technology with which he was unfamiliar. Instead, if the Firm’s Fastcase subscription was too
limited for what was needed, he could have either asked the Firm to purchase additional
resources or taken advantage of the Westlaw access at one of the bar association law libraries.
Similarly, Mr. Schwartz recognizes that as soon as the ChatGPT cases were questioned, he
should have tried to verify the accuracy of the cases from an independent source. But again,
because he believed that ChatGPT was acting as a reputable search engine, and could not
conceive it would simply make up entire cases and citations, he did not immediately question the
very foundation of how the technology worked. Although this may evidence poor judgment, it
does not rise to the level of subjective bad faith or willful blindness. See New V & J Produce
Corp. v. NYCCaterers Inc., No. 13 CIV. 4861 ER, 2014 WL 5026157, at *6, *8 (S.D.N.Y. Sept.
29, 2014) (declining to impose Rule 11 sanctions under the less-stringent “objective
reasonableness” standard when the attorney’s conduct was “certainly deficient” but “based on [a]
good faith, if mistaken belief”); see also Teamsters, 948 F.2d at 1344 (cautioning that the court
“must not allow hindsight to skew its judgment.”).
B.

Mr. Schwartz Promptly Acknowledged His Mistake

Immediate acceptance of responsibility is also inconsistent with the notion of a lawyer
acting in bad faith. C.f. Rankin, 293 F.R.D. at 382-83 (issuing sua sponte sanctions for attorney
and noting “because she falsely denies [her misconduct],” it has “become even more serious
because it makes it seem she is ‘unwilling to change . . . despite ample opportunity’ to do so.”).
Here, in response to the First OSC, Mr. Schwartz and Mr. LoDuca promptly acknowledged their
mistakes. Mr. Schwartz took full responsibility for including the ChatGPT Cases in the
17

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 22 of 29

opposition papers and also acknowledged that he should have taken better care to verify the
accuracy of the cases. Mr. Schwartz also admitted that he should have familiarized himself with
ChatGPT and its risks before using it for legal research. Schwartz Decl. ¶¶ 16, 19, 33.
Now, in response to this Order, Mr. Schwartz is providing a more fulsome explanation of
how and why he came to use ChatGPT. He also continues to own up to his mistakes. These
actions are inconsistent with someone who submitted materials to a court in bad faith. C.f.
Rankin, 293 F.R.D. at 382-83.
C.

The Firm Should Not Be Sanctioned

While Rule 11(c)(1) states that “[a]bsent exceptional circumstances, a law firm must be
held jointly responsible for a violation committed by its partner, associate, or employee,” there is
no “violation” of Rule 11 here, so the Firm should not be sanctioned. As we have shown, there
is no evidence that the Firm or any of its members acted with subjective bad faith. Indeed, the
Firm believed that it had the research tools necessary for the assignment here: it did not know
that its access to federal cases on Fastcase had been limited until it was too late, and even then,
when Mr. Schwartz turned to ChatGPT, he did not know it would, in essence, defraud him.
There is no violation of Rule 11 or Section 1927, and no reason for the Court to use its inherent
power to sanction the Firm.
D.

Mr. LoDuca’s Actions Do Not Support a Finding of Sanctions

In considering whether sanctions are warranted, a “court is expected to avoid using the
wisdom of hindsight and should test the signer’s conduct by inquiring what was reasonable to
believe at the time the pleading, motion, or other paper was submitted.” See Rule 11 advisory
committee note to 1983 amendment. As demonstrated above and in Mr. LoDuca’s May 25, 2023
sworn affidavit (DE 32), at the time the papers were submitted to this Court, Mr. LoDuca had a

18

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 23 of 29

good faith basis to believe in the veracity of Mr. Schwartz’s research and his reliance on that
research was reasonable.
The Advisory Committee Note to Rule 11 explains that among the factors for a court to
consider in determining an attorney’s culpability in signing an incorrect pleading is “whether the
pleading, motion, or other paper was based on a plausible view of the law; [and] whether [the
attorney] depended on forwarding counsel or another member of the bar.” Id. Moreover, courts
have acknowledged that it is not unreasonable for an attorney to rely on information or research
provided by one’s colleagues. See Braun, 2015 WL 4389893, at *19 (declining to sanction
associate who, despite drafting and signing the complaint, was “the least at fault” because “he
relied on information that was provided to him by his colleagues.”).
Here, Mr. LoDuca understood that Mr. Schwartz, a seasoned and experienced attorney,
had prepared the Opposition in good faith based on research that he had conducted online. Mr.
LoDuca also believed that the propositions in the Opposition set forth by Mr. Schwartz about the
statute of limitations and the cases cited for that proposition appeared entirely reasonable.10
Accordingly, at the time Mr. LoDuca filed the Opposition, he had a good faith basis to rely on
Mr. Schwartz’s research. Moreover, even after opposing counsel informed the Court that it was
unable to locate certain of the cases, any concern Mr. LoDuca may have had was addressed by
the fact that Mr. Schwartz appeared to have located the vast majority of the cases, and at no time
did Mr. LoDuca have any reason even to suspect that the cases Mr. Schwartz cited did not exist.
Mr. LoDuca played a limited role in the submissions now at issue, but at all times took his role
seriously and conducted himself honestly. He believed in the veracity of his submissions to the

10

Indeed, the underlying positions advanced in Respondents’ opposition papers were non-frivolous as they were not
“foreclosed a priori by binding precedent.” See Fishoff v. Coty Inc., 634 F.3d 647, 654 (2d Cir. 2011).

19

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 24 of 29

Court, and his conduct, therefore, does not support a finding of sanctions. See Gallop v. Cheney,
667 F.3d 226, 227 (2d Cir. 2012) (amended on Feb. 3, 2012) (vacating sanctions imposed on
local counsel who only “served a peripheral and subordinate role.”); see also De La Fuente v.
DCI Telecommunications, Inc., 269 F. Supp. 2d 229, 235 (S.D.N.Y.), aff’d and remanded, 82 F.
App’x 723 (2d Cir. 2003). Given that Mr. LoDuca’s actions were undertaken in complete good
faith, the Firm also should not be subject to sanctions as a result of Mr. LoDuca’s actions.
E.

Mr. Schwartz, Mr. LoDuca, and the Firm Acted Reasonably

Even if this Court somehow decides to eschew the “subjective bad faith” standard of In
re Pennie, 323 F.3d at 87, and examines whether the conduct was “objectively unreasonable” –
and it should not – sanctions are still inappropriate. Looking at all the facts and circumstances, it
was not objectively unreasonable for Mr. Schwartz to turn to ChatGPT or for him to accept and
act on what he believed were accurate search results. Similarly, it was not objectively
unreasonable for him to go back to ChatGPT to verify the accuracy of the cases. This is so for
several reasons.
First, as we have shown above, in February 2023, when Mr. Schwartz conducted his
research, ChatGPT and other artificial intelligence solutions were being widely hailed in the
press as the future of lawyering. See Alger Decl., Ex. A. Much less was being written about its
deep flaws, and particularly about its tendency toward falsehoods. Id. at Ex. C. Nor did
ChatGPT’s website adequately warn about this risk: it said merely that the program “may
occasionally generate incorrect information,” a warning that would be true of most any search
engine or social media site.11 See supra at n. 4.

11

Interestingly, when ChatGPT was asked the same questions earlier this week that Mr. Schwartz asked when he
was preparing the Affirmation in Opposition, ChatGPT’s answers included substantial disclaimers about the

20

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 25 of 29

Second, the search results appeared authentic on their face: the case names, docket
numbers and judges all looked proper. See supra at 8, n. 7. Mr. Schwartz had no reason to
believe they were false. Indeed, when Mr. Schwartz asked ChatGPT to confirm its false results,
it did so unabashedly. See supra at 10.
Third, Mr. Schwartz resorted to ChatGPT only when he realized that the Firm’s Fastcase
account no longer allowed complete federal searches. He did not have time to fully research the
“risks and benefits” of this new technology. See N.Y. Rule of Prof’l Conduct 1.1, Comm. 8.
Indeed, there was no clear body of knowledge that he could call upon at the time to do so. This
technology was essentially brand new, not one that Mr. Schwartz had ever used before, and not
one that he is likely to use again. See R. Simon, Simon’s New York Rules of Professional
Conduct Annotated (2020-21 ed.) 1.1:9 at 66 (“A lawyer is not required to stay current with
every new app and every form of technology. . . “); see id. at 61, quoting COSAC Reporter’s
Explanation for Proposed Comments to Rule 1.1 (2015) (“The proposed Comment [8] does not
require a lawyer to keep up with developments in technology that are not relevant to a lawyer’s
own practice”). Mr. Schwartz called upon this new tool in a pinch, having to research a federal
bankruptcy issue with which he was completely unfamiliar. Obviously in hindsight he used the
wrong tool. This technological failure should not result in sanctions.
In short, Mr. Schwartz is not the first lawyer to be beguiled by a new technology, and he
will not be the last.12 But he may be one of the first (at least as far as publicly reported decisions
appear to show) to encounter a highly-touted product that generated wholly fictitious results.

reliability of the information. See Alger Decl., Ex. D. Those disclaimers did not appear when Mr. Schwartz
conducted his research. Compare Schwartz Decl. Ex. A with Alger Decl., Ex. D.
12

See Judge Herbert B. Dixon Jr. (Ret.), My “Hallucinating” Experience with ChatGPT, Judges’ J., Spring 2023, at
37, https://www.americanbar.org/content/dam/aba/publications/judges_journal/vol62no2-jj2023-tech.pdf.

21

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 26 of 29

Likewise, Mr. LoDuca acted reasonably in relying on Mr. Schwartz. Accordingly, no sanctions
should issue under even the lesser reasonableness standard.
III.

THE NOTARIZATION OF THE APRIL 25 AFFIDAVIT WAS NOT FALSE OR
FRAUDULENT
The Order also directs Mr. Schwartz and the Firm to show cause why sanctions should

not issue based on “the use of a false and fraudulent notarization” in the April 25 Affidavit. (DE
33.) The notarization here was legitimate: the April 25 Affidavit was signed by Mr. LoDuca
before Mr. Schwartz and then notarized by him, with Mr. Schwartz affixing his stamp. The
document’s metadata shows it was printed on April 25, 2023, as the document indicated. See
Schwartz Decl. ¶ 26, Ex. B. Nevertheless, although dated April 25, 2023, the notary signature
section states that it was notarized on the “25th day of January, 2023.” (See DE 29 (emphasis
added).) This was the result of a typographical error. As detailed in the declaration of Mr.
Schwartz, the April 25 Affidavit was generated from a pre-existing document from another case,
so that they could maintain the affidavit formatting. ¶ 26-27. Unfortunately, owing to a clerical
oversight, the month in the notary section of the original document (“January”) was not changed
to April before the document was printed for signature. In addition, neither Mr. Schwartz nor
Mr. LoDuca noticed the error before signing and notarizing the document. Schwartz Decl. ¶ 2627; LoDuca Decl. ¶ 6.13 This clerical error does not evidence intentional or bad faith conduct.
See In re Bees, 562 F.3d 284, 288 (4th Cir. 2009) (noting that an “inadvertent mistake” that
erroneously implied wrong date of agency publication “does not justify Rule 11 sanctions.”).

13

Mr. LoDuca has the original signature page from the April 25 Affidavit and will present it to the Court at the June
8, 2023 hearing, unless the Court would prefer that it be submitted sooner.

22

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 27 of 29

IV.

SANCTIONS ARE UNNECESSARY
It is well-established that the purpose of sanctions is to deter the specific conduct that

gave rise to the sanctions inquiry and that sanctions should be no more punitive than necessary.
See Fed. R. Civ. P. Rule 11(c)(4) (“A sanction imposed under this rule must be limited to what
suffices to deter repetition of the conduct or comparable conduct by others similarly situated.”);
Rankin, 293 F.R.D. at 394 (“The sanctions the Court imposes are to deter the specific misleading
representations . . . that the Court was considering. . .”). Here, any sanction would be
unnecessary for at least two reasons.
First, Respondents have already taken significant remedial measures. Mr. Schwartz has
spent significant time educating himself on the state of AI technology in the legal field and the
need to properly verify any technological tools used to assist with legal practice. He has already
completed one continuing legal education course on technological competence and the use of
artificial intelligence. Schwartz Decl. ¶ 34.
The Firm has also taken prompt action to address this situation. Upon learning of the
Order, the Firm immediately hired outside counsel with expertise in legal ethics and professional
responsibility to help the firm implement a plan to address the issues raised in the Order. In the
short term, the Firm has conducted an assessment of its legal research capabilities and has now
made sure that every lawyer has access to both federal and state court research databases, as well
as the necessary training resources to effectively use those databases. See Corvino Decl. ¶¶ 10,
13-14. The Firm is scheduling a continuing legal education training on the use of technology in
legal practice, which will include a module on the risks of artificial intelligence products and a
lawyer’s duty of technological competence. Id. at ¶ 14. The Firm has also scheduled an internal
training to remind all lawyers and staff about the proper practices for notarizing documents,
including a requirement that the notary and the signer verify the accuracy of the information in
23

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 28 of 29

the notary signature section. Id. at ¶ 15. Accordingly, any sanction for the same conduct would
be duplicative and unnecessary to act as a deterrent. See Rivas, 2014 WL 3694983, at *4.
Second, the public scrutiny to which Respondents have been subjected as a result of this
Court’s Order to Show Cause has also been an effective deterrent both for Respondents and
others similarly situated. In the past week, there have been numerous articles, social media
posts, and other online media about Respondents and this case. See Alger Decl., Ex. E (articles
discussing and/or mentioning this Court’s Order and identifying Mr. Schwartz, Mr. LoDuca, and
the Firm by name). Articles about Respondents have already been circulated in law firms across
the country as a cautionary tale. Id. Respondents – and Mr. Schwartz in particular – have been
publicly shamed for their supposed ignorance and carelessness. Id. Their reputations, and the
Firm brand, have been irreparably damaged. To impose further sanctions on top of this would be
unduly punitive and serve no useful purpose.
CONCLUSION
For the reasons stated above and in the accompanying Declarations, this Court should not
impose sanctions on Mr. Schwartz or the Firm pursuant to Rule 11, 28 U.S.C. § 1927 or its
inherent power.
Dated: New York, New York
June 6, 2023
FRANKFURT KURNIT KLEIN & SELZ, P.C.
By:

/s/ Tyler Maulsby
Tyler Maulsby
Ronald C. Minkoff
Ashley K. Alger

28 Liberty Street, 35th Floor
New York, New York 10005
Tel.: [phone]
Fax: [phone]
[e-mail]
24

Case 1:22-cv-01461-PKC Document 45 Filed 06/06/23 Page 29 of 29

[e-mail]
[e-mail]
Attorneys for Non-Party Stephen A. Schwartz, Esq.
and Levidow Levidow & Oberman, P.C.

25