Opinion

Garner

Court: District Court, W.D. Washington
Filed: Mar 30, 2026
Cited by: 0 cases
Authority: More cited than 41.4%

The opinion

UNITED STATES DISTRICT COURT

FOR THE WESTERN DISTRICT OF WASHINGTON

6 AT SEATTLE

KAELI GARNER, et al.,

8 CASE NO. 2:21-cv-00750-RSL

Plaintiffs,

10 ORDER GRANTING IN PART

AMAZON.COM, INC., et al., DEFENDANTS’ DAUBERT MOTION

11 REGARDING TESTIMONY OF

SERGE EGELMAN

Defendants.

This matter comes before the Court on “Amazon’s Motion to Exclude Testimony of

Serge Egelman.” Dkt. # 322. Defendants seek to exclude Dr. Egelman’s testimony

18 regarding (a) whether voice recordings are personally identifiable information, (b) the

frequency and content of false wake recordings, (c) the reliability of voice recordings,

(d) consumer expectations and views regarding the retention and use of voice recordings,

(e) whether users consented to Alexa’s terms and conditions, (f) the content of Amazon’s

23 documents, and (g) Dr. Egelman’s interpretation of Amazon’s documents.1

Federal Rule of Evidence 702 provides that expert testimony is admissible if:

1 This matter can be decided on the papers submitted. Amazon’s request for oral argument is DENIED.

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 (1) the witness is sufficiently qualified as an expert by knowledge, skill,

experience, training, or education; (2) the scientific, technical, or other

specialized knowledge will help the trier of fact to understand the evidence

3 or to determine a fact in issue; (3) the testimony is based on sufficient facts

or data; (4) the testimony is the product of reliable principles and methods;

4 and (5) the expert has reliably applied the relevant principles and methods to

5 the facts of the case.

6 City of Pomona v. SQM N. Am. Corp., 750 F.3d 1036, 1043 (9th Cir. 2014). As construed

in Daubert v. Merrell Dow Pharmaceuticals, Inc., Rule 702 tasks a district judge with

“ensuring that an expert’s testimony both rests on a reliable foundation and is relevant to

10 the task at hand.” 509 U.S. 579, 597 (1993). Where an expert offers non-scientific

11 testimony, “reliability depends heavily on the knowledge and experience of the expert,

rather than the methodology or theory behind” the testimony. Porter v. Martinez, 64 F.4th

1112, 1127 (9th Cir. 2023) (quoting Daubert, 509 U.S. at 594, and Hangarter v. Provident

15 Life & Acc. Ins. Co., 373 F. 3d 998, 1017 (9th Cir. 2004)). The analysis “should be applied

16 with a ‘liberal thrust’ favoring admission.” Messick v. Novartis Pharms. Corp., 747 F.3d

1193, 1196 (9th Cir. 2014) (quoting Daubert, 509 U.S. at 588).

Ultimately, the test under Daubert is not the correctness of the expert’s

19 conclusions but the soundness of his methodology. The court is a gatekeeper,

not a fact finder. Accordingly, the district court is not tasked with deciding

whether the expert is right or wrong, just whether his testimony has

21 substance such that it would be helpful to a jury. If the proposed testimony

meets the thresholds of relevance and reliability, its proponent is entitled to

have the jury decide upon its credibility, rather than the judge. Challenges

23 that go to the weight of the evidence are within the province of a fact finder,

not a trial court judge. A district court should not make credibility

24 determinations that are reserved for the jury. This Court has previously noted

that shaky but admissible evidence is to be attacked by cross examination,

contrary evidence, and attention to the burden of proof, not exclusion.

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 Elosu v. Middlefork Ranch Inc., 26 F.4th 1017, 1024 (9th Cir. 2022) (internal quotation

marks, citations, and alterations omitted). “Basically, the judge is supposed to screen the

jury from unreliable nonsense opinions, but not exclude opinions merely because they are

5 impeachable.” Alaska Rent-A-Car, Inc. v. Avis Budget Grp., Inc., 738 F.3d 960, 969–70

6 (9th Cir. 2013).

A. Voice Recordings as Personally Identifiable Information

Dr. Egelman opines that voice recordings are personally identifiable information.

10 Dkt. 324 at ¶¶ 47-51. He explains why voices are unique and how voice recordings can be

11 matched, and he provides examples in research, statutes, and within Amazon of voice

recordings being treated as personally identifiable information. Amazon does not take

issue with that opinion, per se, but rather argues that plaintiffs have improperly relied “on

15 that generalized statement to suggest that Amazon could identify all people whose voices

16 were recorded in Alexa interactions.” The portion of plaintiffs’ motion for class

certification to which Amazon apparently objects relies on evidence other than Dr.

Egelman’s report, however. Dkt. 257 at 18 (“Moreover, Amazon has a voice profile for

Non-Registrants tied to the IP address and other identifying data associated with the

21 account. See Ex. CC at 303:12-24; Ex. WW at ‘7633.”). Even if, as Amazon asserts,

plaintiffs stretched Dr. Egelman’s general statement to support a broader argument,

counsels’ advocacy does not make unreliable or unhelpful Dr. Egelman’s opinion

regarding the personally identifiable nature of voice recordings.

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 B. Frequency and Content of False Wakes

Dr. Egelman, citing an internal Amazon document, opines that Alexa devices

experience between 0.21 and 0.43 false wakes per week and that “each and every Alexa

5 device will inappropriately intercept and transmit contemporaneously-captured audio

6 streams at least once per month.” Dkt. 324 at ¶ 18 (emphasis in original). Amazon takes

issue with this opinion, asserting that the document on which Dr. Egelman relies reflects

the average rate of false wakes per week and cannot logically be used to say anything

10 about the number of false wakes a particular device might experience. Plaintiffs, on the

11 other hand, assert that the document discusses actual false wake measurements. A review

of AMZ_GARNER_00052921 at ‘2921 does not resolve the matter: the top end of the

false wake range appears to be measured on a device/week basis (i.e., actual rates) while

15 the bottom end of the range is clearly stated as an average. Dkt. 379-2 at 3. Even if all of

16 the numbers in the range are based on averages, Dr. Egelman maintains that the complex

statistical models generating these numbers, the way the technology works, and the

millions of data points lead to the conclusion that all of the devices are impacted by false

wakes. Dkt. 379-1 at 5-7. While this conclusion is not the only one that could be drawn

21 from the underlying data, “[s]haky but admissible evidence is to be attacked by cross

examination, contrary evidence, and attention to the burden of proof, not exclusion.”

Primiano v. Cook, 598 F.3d 558, 564 (9th Cir. 2010), as amended (Apr. 27, 2010).

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 C. Consumer Awareness of Alexa’s Recording Practices

In a section of his report entitled “The Public is Likely Unaware Audio was

Collected,” Dr. Egelman makes a number of statements regarding how Alexa and its opt-

5 out features work and cites research studies showing that “almost half” of Alexa users did

6 not know their recordings were being permanently stored or that they could review them

and that “most participants” did not feel like they had consented to Amazon’s storage and

use of their data. Dkt. 324 at ¶¶ 36-37. Amazon combines the title, the technological

10 descriptions, and the studies to ascribe to Dr. Egelman opinions he has not espoused. Dkt.

11 322 at 5-6 (attributing to Dr. Egelman Amazon’s descriptions of the product and the

statement of an anonymous study participant); Dkt. 379 at 7 (attributing to Dr. Egelman

the opinion that customers “are uniformly unaware” of how Alexa works). To the extent

15 Dr. Egelman is opining that the public is likely unaware of Alexa’s recording, collection,

16 storage, and review features, the opinion is based on sufficient data and facts, including

relevant studies and his two decades of experience/expertise in researching consumer

privacy preferences and expectations. The fact that Amazon can identify additional sources

of facts and data that were not considered goes to the weight of the evidence, not its

21 admissibility.

D. Consumer Views of Alexa’s Recording Practices

Dr. Egelman opines that, based on his prior research regarding consumer privacy

and security decision making, consumer privacy preferences, privacy and security

26 expectations, and the ways online services and other software attempt to satisfy those

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 expectations, consumers would find Alexa’s recording, collection, storage, and review

features objectionable. In particular, Dr. Egelman relies on studies regarding consumers’

reactions to recording within their homes, the personally identifiable nature of voice

5 recordings, consumer reactions to sharing recordings with humans (as opposed to machine

6 processing), and the strenuous objections of “bystanders” (those who are not users of the

device but nevertheless have their data captured), as well as Amazon’s own internal

acknowledgment that using consumer voice recordings raises privacy concerns for Alexa

10 users. Amazon argues that this opinion is baseless because the technologies that were

11 studied (smartphones, wearable devices, etc.) differ from voice assistants like Alexa and

the consumers participating in the study do not map squarely with any of the classes

proposed by plaintiffs. It further asserts that Dr. Egelman misinterprets or misstates the

15 findings of three of the studies on which he relies.

16 Dr. Egelman has sufficient knowledge, experience, and data to opine regarding how

consumers would view various recording practices and uses of their voice data. Some of

the surveys on which Dr. Egelman relies relate to voice assistants like Alexa, and even

studies that were not specific to voice assistant technologies shed light on how consumers

21 view the collection and use of their data. The survey results that Amazon believes

contradict Dr. Egelman’s opinions are, on the whole, supportive. The Malkin, et al. (2018)

study, for example, amply supports the conclusion that, while between 25-50% of study

participants recognized that data collected by internet-enabled devices could be shared and

26 repurposed, a large majority strongly felt that data sharing was unacceptable and would, if

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 given an opportunity, disable any data-sharing features. Dkt. 325-3 at 103-05. Again, while

Dr. Egelman’s conclusions regarding consumer views are not unassailable, “[s]haky but

admissible evidence is to be attacked by cross examination, contrary evidence, and

5 attention to the burden of proof, not exclusion.” Primiano, 598 F.3d at 564.

6 E. Consent

Dr. Egelman opines that the collector of personal data has an obligation “to ensure

customers are informed and consent to the collection of the data and how it may be used

10 (e.g., that it may be shared with human employees/contractors who will listen to it) –

11 which Amazon failed to do here" and that non-registrants did not have the opportunity to

consent. Dkt. 324 at 14-15. Amazon argues that any opinion regarding whether users

expressly or impliedly consented to Alexa recordings must be excluded as an improper

15 legal conclusion, is without basis, and is contrary to the Court’s prior rulings.

16 Amazon’s underlying assertion that consent is a legal issue to be decided by the

Court is unsupported. While a Court may resolve the issue of consent if there is no genuine

dispute regarding a material fact, there are myriad cases and contexts in which consent is

deemed a fact issue. See, e.g., Harrison v. Eddy Potash, Inc., 248 F.3d 1014, 1023 (10th

21 Cir. 2001) (consent to sexual advances); United States v. Buettner-Janusch, 646 F.2d 759,

764 (2d Cir. 1981) (consent to search); Godinez-Torres v. United States, No. 14-cv-01097-

CBA-PK, 2016 WL 11670284, at *10 (E.D.N.Y. Mar. 31, 2016) (consent to surgery);

Anderson v. City of Columbus, Georgia, 374 F. Supp. 2d 1240, 1250 (M.D. Ga. 2005)

26 (consent to recording). Nevertheless, Dr. Egelman will not be permitted to opine regarding

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 the consent (or lack of consent) of consumers as a whole or of individual consumers. Dr.

Egelman’s expertise is in consumer expectations and preferences. He may, as discussed

above, opine regarding what consumers expect with regards to the collection and use of

5 their data, what their preferences would be, and what Amazon could have or should have

6 done to ensure that those expectations and preferences were satisfied. But neither plaintiffs

nor Dr. Egelman establish any basis for an opinion regarding whether a particular

individual consumer or consumers as a group consented to the collection of audio

10 recordings or the storage, review, and subsequent use of the same.

11 F. Contents of Amazon Documents

Amazon seeks to exclude Dr. Egelman’s statements regarding how Alexa works

and what Amazon does with the resulting voice recordings on the ground that the

15 statements merely regurgitate the content of Amazon’s documents without analyzing or

16 synthesizing the information into admissible expert opinions. An explanation of the

product and its capabilities is necessary background for Dr. Egelman’s opinions. For

example, knowing that the technology generates unintended recordings and what Amazon

does with those recordings helps establish the relevance of the research on which Dr.

21 Egelman relies to opine on consumer expectations and preferences. While Dr. Egelman

will be instructed at trial to differentiate between facts he has assumed based on Amazon’s

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 documents2 and the opinions he has drawn from those facts, he is not precluded from

discussing Amazon’s documents in this context.

G. Interpretation of Amazon’s Documents

5 Amazon seeks to exclude what it describes as Dr. Egelman’s incorrect

6 interpretations and misleading characterizations of Amazon’s documents. In paragraph 16,

for example, Dr. Egelman apparently interpreted an on-going discussion regarding whether

the “comms team” stored “data in any place other than Datamart” (Dep. Tr. At 197) as an

10 admission that “Amazon is not sure of all the places where employees and/or contractors

11 store voice data” (Dkt. 324 at ¶ 16).

The Court takes this matter under advisement. Dr. Egelman’s expertise lies in

evaluating consumer expectations, consumer preferences, and the ways in which products

15 meet those expectations/preferences. To the extent Dr. Egelman relies on an Amazon

16 document to support an opinion that falls within his expertise, his understanding of the

document is admissible even if faulty. But there is no reason to believe that Dr. Egelman

has any special knowledge or experience that would allow him to interpret or explain the

meaning of internal Amazon documents that have nothing to do with the opinions offered.

21 The admissibility of the challenged portions of the report will be determined based on the

context in which the testimony is elicited at trial.

2 The underlying facts regarding how Alexa works and what Amazon does with the resulting data must, of course,

26 be established through the presentation of admissible evidence. To the extent Dr. Egelman’s assumed facts are

incorrect, that can be shown on cross-examination.

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

1 For all of the forgoing reasons, defendants’ Daubert motion regarding Dr.

Egelman’s opinions is GRANTED in part. Dr. Egelman will not be permitted to opine

regarding whether a particular consumer or group of consumers consented to Amazon’s

5 collection of audio recordings or their storage, review, and subsequent use. Amazon’s

6 objection to Dr. Egelman’s interpretations of its internal documents is taken under

advisement. The motion to exclude is DENIED in all other respects.

Dated this 30th day of March, 2026.

12 Robert S. Lasnik

United States District Judge

ORDER GRANTING IN PART DEFENDANTS’ DAUBERT

MOTION REGARDING TESTIMONY OF SERGE

This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.