Jefferies et al v. Harcros Chemicals Inc. et al

1
IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF KANSAS
ELLSWORTH WILLIAM JEFFRIES, III, et al.,
Plaintiffs,
v.
HARCROS CHEMICALS INC., et al.,
Defendants.
Case No. 25-2352-KHV-ADM
VANESSA TUCKER, individually and on behalf
of those similarly situated,
Plaintiff,
v.
HARCROS CHEMICALS INC., et al.,
Defendants.
Case No. 25-2569-KHV-ADM
MEMORANDUM AND ORDER
These consolidated cases are putative class actions brought by plaintiffs who live within a
three-mile radius of a chemical facility in Kansas City, Kansas. Plaintiffs contend this facility has
emitted a “toxic cocktail” of emissions that have poisoned and contaminated the air, leading to
severe and chronic injuries to residents in the surrounding vicinity. Defendants are a series of
entities that collectively owned and operated the facility over a period of several decades, including
Harcros Chemicals, Inc. (“Harcros”), Philips North America LLC f/k/a Philips Electronics North
America Corporation and Koninklijke Phillips N.V. (“Philips”), and Elementis Chemicals, Inc.
(“Elementis”). This case is now before the court on Defendants’ Motion to Amend the Protective
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 1 of 9

2
Order that the court has already entered to govern discovery in this case. (ECF 120.) For the
reasons explained below, Defendants’ motion is granted.
I. BACKGROUND
When the court originally entered the protective order in this case (ECF 115), the court
resolved the parties’ first round of disputes over restrictions for using AI Tools in relation to
Confidential Information. The protective order provides, among other things, that a party
intending to use a particular AI Tool must provide other parties with notice and an opportunity to
object; ensure that the AI Tool is used in a secure environment and that Confidential Information
is not used to train or improve any AI Tool except one used exclusively in this action; to the extent
that an AI Tool is trained or improved using Confidential Information, that information is
destroyed at the conclusion of this litigation and is not made accessible to anyone not authorized
to have access to Confidential Information; and ensure that all Confidential Information is deleted
at the conclusion of this action. (Id. ¶ 8, at 6-8.) As a practical matter, these provisions effectively
require parties to use only closed or secure AI Tools (“closed AI Tools”) for Confidential
Information. At the time the court entered the protective order, the parties foreshadowed a broader
dispute over the use of AI Tools in connection with all information produced in discovery in this
case, not just Confidential Information.
The parties have further crystallized their dispute on this point, which is now before the
court on Defendants’ Motion to Amend the Protective Order. (ECF 120.) By way of this motion,
Defendants seek to amend the protective order language to expand the AI provisions described
above to apply not only to documents and information designated as “Confidential,” but more
broadly to all “Discovery Materials”—i.e., all documents and information produced in discovery.
Highly summarized, Defendants seek to prevent parties from uploading even non-confidential
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 2 of 9

3
documents into public or “open loop” generative artificial intelligence tools (collectively, “open
AI Tools”).
For the reasons stated below, the court finds good cause to enter the amended version of
the protective order proposed by Defendants. The court therefore grants Defendants’ motion and
enters an Amended Protective Order in line with the proposed protective order attached to
Defendants’ motion (ECF 120-1).
II. LEGAL STANDARD
Under Federal Rule of Civil Procedure 26(c)(1), the court may, for good cause, “issue an
order to protect a party or person from annoyance, embarrassment, oppression, or undue burden
or expense.” Rule 26(c)(1) permits a court to issue a protective order “limiting the scope of
disclosure or discovery to certain matters.” See FED. R. CIV. P. 26(c)(1)(D). Rule 26(c)(1)’s good-
cause standard is “highly flexible, having been designed to accommodate all relevant interests as
they arise.” Rohrbough v. Harris, 549 F.3d 1313, 1321 (10th Cir. 2008).
III. ANALYSIS
Defendants assert that good cause exists for their proposed amendment governing the
disclosure of non-confidential materials because the availability and use of AI Tools in litigation
“represents a paradigm shift of yet-to-be defined proportions, but what is clear is that the way AI
Tools function poses a potential threat to the integrity and security of data produced in litigation.”
(ECF 120, at 2.) Defendants explain that AI Tools rely on sophisticated machine learning models
that work by identifying and encoding patterns and relationships in massive amounts of data and
then using that information to understand users’ natural language requests or questions and respond
with relevant new content. But, unlike closed AI Tools, the use of open AI Tools “risks disclosure,
loss of control, and uncertainty concerning the security, storage, and other data-handling of that
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 3 of 9

4
information.” (Id. at 2-3.) Defendants point out that, because an open AI Tool uses the data
submitted to it to continually develop and improve the tool, it is “practically impossible” to claw
back data later determined to be privileged, or delete data from the open AI Tool at the conclusion
of the action (as required by the deletion clause in the Protective Order) because the information
was used to train the AI Tool. (Id. at 3, 6-7.) Defendants further explain the harm that might result
if information cannot be clawed back or deleted—e.g., waiving confidentiality or privilege of
information that was inadvertently disclosed without a “confidential” designation; revealing
sensitive data or personal information from the AI Tool’s training datasets; and potentially
violating counsel’s professional duty to safeguard information relating to representation of a client.
(Id. at 3-5.) Defendants also point out that wholesale submission of discovery materials to an open
AI Tool may violate U.S. data privacy laws and the strict disclosure rules under the European
General Data Protection Regulation (“GDPR”), to which defendants Elementis and Philips are
subject. (Id. at 7-8.) Lastly, Defendants contend that their proposed amendment is necessary to
protect against exposure of critical infrastructure and data breach, given that Defendants are part
of the chemical sector designated as Critical Infrastructure and vital to national security. (Id. at 8-
9.) Defendants contend that all of these concerns constitute good cause to amend the protective
order.
Plaintiffs dispute that Defendants have shown good cause for what Plaintiffs characterize
as an “umbrella” protective order that is disfavored by courts. Plaintiffs also contend that
Defendants’ proposal would drive up Plaintiffs’ costs by denying them access to open AI Tools to
analyze “otherwise unprotected discovery materials.” (ECF 123, at 5-6, 8.) Plaintiffs further argue
that Defendants’ proposal deprives Plaintiffs of their First Amendment right to use and disseminate
nonconfidential discovery materials to public-use, open AI Tools. (Id. at 4, 5.) Lastly, Plaintiffs
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 4 of 9

5
say that Defendants’ arguments regarding the impossibility of clawing back or deleting the
documents submitted to an open AI Tool and regarding potential violations of privacy laws and
the GDPR are only speculative. (Id. at 8-9.)
To begin, the court disagrees with Plaintiffs’ characterization of Defendants’ proposal as a
disfavored “umbrella” protective order. Umbrella protective orders preemptively designate all
discovery as protected without any review by either the court or the parties. See Mann v. Con-
Way Freight, No. 16-2196-CM-JPO, 2016 WL 6164013, at *4 (D. Kan. Oct. 24, 2016) (citation
omitted). Although Defendants’ proposed protective order covers all discovery materials, the
parties will still be reviewing and screening discovery materials to determine the extent to which
they should be marked “Confidential.” So, Defendants are not proposing the type of “umbrella”
protective order that is disfavored by the courts.
The court turns next to Plaintiffs’ contention that Defendants’ proposal seeks to increase
Plaintiffs’ litigation costs by depriving them of open AI Tools to analyze discovery materials and
making it more burdensome for Plaintiffs to pursue this litigation. (ECF 123, at 5.) Plaintiffs have
not presented any support for the increased burden, such as by attempting to quantify the extent of
any such increased cost. Thus, the court does not find that Plaintiffs would suffer an undue burden
if the court were to grant Defendants’ motion. Indeed, Plaintiffs and their attorneys initiated this
lawsuit knowing that class action litigation is expensive, including the costs of eDiscovery, expert
witness fees, and administrative, court-filing, and other expenses. And, as Defendants point out,
their proposal “applies equally to all parties and allows all parties to use Closed AI Tools that meet
basic security requirements,” thereby putting the same financial burden on Plaintiffs and
Defendants. (ECF 129, at 1.)
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 5 of 9

6
Next, the court turns to Plaintiffs’ contention that Defendants’ proposed amended
protective order deprives Plaintiffs of their First Amendment right to use and disseminate
nonconfidential discovery materials. Not so. A protective order does not offend the First
Amendment when it is entered based on a showing of good cause, is limited to the context of
pretrial civil discovery, and does not restrict the dissemination of the information if gained from
other sources. Seattle Times Co. v. Rhinehart, 467 U.S. 20, 37 (1984). Here, Defendants’ proposed
amended protective order does not prohibit the parties from disseminating non-confidential
documents to the public. Indeed, Defendants concede that their proposed amended protective
order “does not in any respect affect how all parties—consistent with applicable law, ethical
requirements, and professional standards—can publicize non-confidential documents by posting
such documents to a website, making documents available to the media, or attaching documents
to a court filing.” (ECF 129, at 2-3.)
Lastly, the court addresses Plaintiffs’ argument that Defendants have not met their burden
to show the need for the relief sought. Plaintiffs point out that Defendants did not submit an
affidavit from a cybersecurity expert to back up their claims of risk or to analyze the regulatory
framework of the GDPR or provide specific examples of problematic discovery materials.
Plaintiffs view Defendants’ proposal as seeking to “convert all discovery into protected material .
. . based on speculative fears untethered to any concrete showing of harm.” (ECF 123, at 4, 5, 8-
9.) In reply, Defendants point out that their motion presents an “issue of first impression” and call
attention to the documents and policies they cited in their motion, which “establish the scrutiny
and care that must be taken when submitting data, particularly that from within a critical
infrastructure industry, to Public AI Tools” given the “AI-specific security risks that may expose
training or other data.” (ECF 129, at 3-4.)
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 6 of 9

7
On balance, Defendants have the better arguments. Plaintiffs never directly address
Defendants’ argument that clawing back or deleting data submitted to an open AI Tool is
impossible as a practical matter because such data is used to continually develop and improve the
tool. Nor do they address Defendants’ argument that the use of an open AI Tool may expose
Defendants’ critical infrastructure information to cyber criminals and risk data breach. Instead,
they just call the arguments “speculative” and insist that any resulting harm from the information’s
disclosure is no greater than if Plaintiffs publicly posted the nonconfidential information on a
website, which the current version of the protective order would not prohibit. (ECF 123, at 8-9.)
But this analogy is not persuasive. The use of widely available AI Tools presents the opportunity
for a centralized repository that makes information available to the public at a scale that was not
historically available and ignores the very real security risks of public AI Tools, including the
inability to effectively claw back information from the AI Tool.
It is indisputable that generative AI may automate many time-consuming tasks in the
eDiscovery process, such as significantly accelerating document review, quickly summarizing
documents, streamlining privilege review and logging, identifying named entities like key people
and organizations, extracting important topics, and automating writing tasks, such as drafting
deposition questions, creating narrative timelines from evidence, and drafting legal briefs. But
Defendants’ proposal does not foreclose a party from using any AI Tools; it only prohibits using
open AI Tools while allowing the use of closed AI Tools—for good reason. The wholesale
submission of discovery materials to an open AI Tool for these eDiscovery tasks could expose
massive amounts of data. These actions may violate U.S. data privacy laws and the stricter GDPR
disclosure rules, which set a high standard for protecting individuals’ information and requires
documented consent to be “freely given, specific, informed and unambiguous.” (GDPR Article
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 7 of 9

8
4(11).) As Defendants point out, Defendants’ employees, contractors, and correspondents have
not consented. Thus, Plaintiffs’ proposal, which proposes only redaction of certain personally
identifiable information, does not appear to comply with the strict requirements of the GDPR,
including the consent requirement.
On balance, the court finds that Defendants have met their burden to show the need for the
relief sought given the AI-specific risks identified by Defendants in this particular case—namely,
where documents to be submitted to the AI Tool come from entities in a critical infrastructure
industry and disclosure may violate data privacy laws. Plaintiffs’ arguments do not persuade the
court otherwise. Parties are often required to produce massive amounts of information in complex
litigation, and this often includes documents that may include a small amount of information that
is responsive to discovery requests amongst otherwise largely irrelevant or non-responsive
information. See, e.g., Ritchie v. Wal-Mart Stores East, L.P., No. 19-4067-SAC-ADM, 2020 WL
5369392, at *2 (D. Kan. Sept. 8, 2020) (“Ordinarily, a party may not unilaterally redact
information from documents it produces in litigation simply because the redacted information is
irrelevant or non-responsive.”). If the court were to allow a party to upload all non-confidential
documents and materials produced by another party to an open AI Tool, thereby making all such
data amenable to public consumption, parties may err on the side of under-producing potentially
responsive documents or seek to make extensive redactions of irrelevant or non-responsive
information. Defendants’ proposed amendment allowing the use of closed AI Tools, as opposed
to open AI Tools, will facilitate discovery by incentivizing more fulsome document productions.
IV. CONCLUSION
In conclusion, the court finds that Defendants have shown good cause for entry of
Defendants’ proposed amended protective order for the reasons stated above.
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 8 of 9

9
IT IS THEREFORE ORDERED that Defendants’ Motion to Amend the Protective
Order (ECF 120) is granted. The court will enter the Amended Protective Order forthwith.
IT IS SO ORDERED.
Dated March 25, 2026, at Kansas City, Kansas.
s/ Angel D. Mitchell
Angel D. Mitchell
U.S. Magistrate Judge
Case 2:25-cv-02352-KHV-ADM Document 152 Filed 03/25/26 Page 9 of 9