Court Decision on Amendment Request in Elon Musk's X Breach of Contract Lawsuit

3. Amendment

The Court must next decide whether to allow X Corp. leave to amend its breach of contract claim. The Court is mindful that “[t]he anti-SLAPP statute was enacted to allow early dismissal of meritless first amendment cases aimed at chilling expression through costly, time-consuming litigation,” Metabolife Int’l, Inc. v. Wornick, 264 F.3d 832, 839 (9th Cir. 2001), but also that “granting a defendant’s anti-SLAPP motion to strike a plaintiff’s initial complaint without granting the plaintiff leave to amend would directly collide with Fed. R. Civ. P. 15(a)’s policy favoring liberal amendment,” Verizon Delaware, Inc. v. Covad Commc’ns Co., 377 F.3d 1081, 1091 (9th Cir. 2004). To be clear, the operative complaint here is not X Corp.’s “initial complaint”—X Corp. initially filed suit on July 31, 2023, see Compl., and amended about a week later, adding ECF as a defendant, see FAC. Nevertheless, the Court agrees with X Corp. that it “is not as if [the defendants] have been dragged through multiple, multiple rounds” of amendments. See Tr. of 2/29/24 Hearing at 34:8–18;[20] but see Gardner v. Martino, 563 F.3d 981, 991 (9th Cir. 2009) (“Appellants had already filed the first amended complaint as a matter of right and Verizon specifically held that a first amended complaint is subject to anti-SLAPP remedies.”) (citing Verizon Del., Inc., 377 F.3d at 1091).

The question is whether X Corp. should receive leave to amend under Rule 15. See Mahoney v. Meta Platforms, Inc., No. 22-cv-02873-AMO, 2024 WL 68550, at *5–6 (N.D. Cal. Jan. 6, 2024) (holding that plaintiff failed to state a claim, but because amendment was “not clearly futile,” it was appropriate to defer consideration of defendant’s antiSLAPP motion pending filing of amended complaint). A court should “freely give leave” to amend “when justice so requires.” Fed. R. Civ. Proc. 15(a)(2). It may deny leave to amend due to “undue delay, bad faith or dilatory motive on the part of the movant, repeated failure to cure deficiencies by amendment previously allowed, undue prejudice to the opposing party by virtue of allowance of the amendment, [and] futility of amendment.” Leadsinger, 512 F.3d at 532.

The Court believes that amendment would be futile, and that X Corp.’s desire to amend may well be based on a dilatory motive.

At the motion hearing, X Corp. argued for the first time that there were damages that it had “not pleaded in the complaint,” but could add upon amendment, “which [are] in direct response to this scraping incident.” Tr. of 2/29/24 Hearing at 8:18–20. X Corp. described these damages as “dozens, if not over a hundred personnel hours across disciplines such as engineering and security that were spent trying to figure out not only what data was gained via access to the Brandwatch systems”—time not relevant to the breach of contract claim—“but what data was scraped, how it was done.” Id. at 8:7–11. It further referenced “anti-scraping measures implemented, again, at great expense from engineering time in tens of thousands, if not over a hundred thousand dollars to deal with this security risk that was identified via this incident.” Id. at 8:21–24; see also id. at 9:3–6 (“I’m sure we had, from what I understand, cost for the servers dealing with the load of just receiving all these requests. So expensive electricity cost at data centers, as well as loss of computing power.”).

Damages stemming from X Corp.’s response to CCDH’s alleged scraping—rather than advertisers’ responses to CCDH’s reports—do not present the same constitutional concerns that the current damage allegations do. But they must still have been “within the contemplation of the parties when the contract was entered into or at least reasonably foreseeable by them at that time,” under state contract law. See Erlich, 21 Cal. 4th at 550. The Court is unpersuaded that the proposed allegations clear that hurdle. Fundamentally, neither of the two main reasons that X Corp. asserted for having to spend time and money responding to CCDH’s alleged scraping—(a) “safety and security” and (b) harm to X Corp.—make very much sense.

a. Safety and Security

First, X Corp. argued that it needed to expend resources responding to CCDH’s alleged breach in order to safeguard its users’ data. Over and over again, X Corp. played up its obligation to “keep [its users’] data safe,” Tr. of 2/29/24 Hearing at 7:13–14, called scraping a “security risk,” id. at 8:23–24, and declared that this case was “about” “the importance of data security,” id. at 12:10–11; see also id. at 14:9–10 (arguing that the CCDH report acknowledging scraping “raised a potential security issue”); id. at 35:13–15 (“we put up anti-scraping measures to stop this kind of thing so that users feel as if their data is safe.”); id. at 39:22–24 (“This is about the security of data and whether or not we’re going to honor ways that people try to keep it safe contractually and otherwise.”).

While security and safety are noble concepts, they have nothing to do with this case. The Toxic Twitter report stated that CCDH had used the SNScrape tool, “which utilizes Twitter’s search function,” to “gather tweets from” “ten reinstated accounts,” resulting in a “dataset of 9,615 tweets posted by the accounts.” Toxic Twitter at 17. There is no allegation in the complaint, and X Corp. did not assert that it could add an allegation, that CCDH scraped anything other than public tweets that ten X platform users deliberately broadcast to the world.[21] No private user information was involved—no social security numbers, no account balances, no account numbers, no passwords, not even “gender, relationship status, ad interests etc.” See Meta Platforms, Inc. v. BrandTotal Ltd., 605 F. Supp. 3d 1218, 1273 (2022).

When asked why the collecting of public tweets implicates users’ security interests, X Corp. insisted that “this all goes to control over your data,” and that users expect that they will be able to take down their tweets later, or to change them—abilities they are robbed of when the “data” is scraped. See Tr. of 2/29/24 Hearing at 40:17–24. But even assuming that it is “very important” to a user that he be able to amend or remove his proneo-Nazi tweets at some point after he has tweeted them, see id. at 40:24–25, a user can have no expectation that a tweet that he has publicly disseminated will not be seen by the public before that user has a chance to amend or remove it. While scraping is one way to collect a user’s tweets, members of the public could follow that user and view his tweets in their feeds, or use the X platform’s search tool (as SNScrape did) and view his tweets that way.

X Corp.’s assertion that the scraping alleged here violates a user’s “safety and security” in his publicly disseminated tweets is therefore a non-starter. CCDH would not, and should not, have foreseen at the time of contracting that X Corp. would need to expend money to address “safety and security” issues following the scraping of ten users’ public tweets.

b. Harm to X Corp.

Second, X Corp. argued that it was entitled to expend resources responding to CCDH’s scraping because that scraping might have caused harm to X Corp.’s systems. X Corp. maintained that it was entitled to “dive in and understand what was done, how it was done, what data was taken, whether there was any damage,” and to conduct “a large investigation and then remedial action,” Tr. of 2/29/24 Hearing at 9:16–20, adding that “[i]t is well known in this space that if you scrape,” the scrapee will need to investigate and “put up anti-scraping measures,” id. at 35:9–14. X Corp. conceded that it is not able to prevent scraping “100 percent” of the time, but stated that it can make scraping more difficult, and should be compensated for trying to do so. Id. at 35:21–36:7.[22] The problem with this argument is that it is at odds with what X Corp. has alleged.

Although social media platforms do not like it, scraping, for various ends, is commonplace. See, e.g., ACLU Br. at 7 (“Researchers and Journalists Use Scraping to Enable Speech in the Public Interest and Hold Power to Account.”); see also id. at 8–9 (collecting sources); Andrew Sellars, “Twenty Years of Web Scraping and the Computer Fraud and Abuse Act,” 24 B.U. J. Sci. & Tech. L 372, 375 (2018) (“Web scraping has proliferated beneath the shadow of the [CFAA].”); hiQ 2022 Circuit opinion, 31 F.4th at 1202 (“HiQ points out that data scraping is a common method of gathering information, used by search engines, academic researchers, and many others. According to hiQ, letting established entities that already have accumulated large user data sets decide who can scrape that data from otherwise public websites gives those entities outsized control over how such data may be put to use . . . the public interest favors hiQ’s position.”); see also id. at 1186 (“LinkedIn blocks approximately 95 million automated attempts to scrape data every day”).

Some scraping activity can be harmful not only to the users of a website but also to the website itself. See, e.g., hiQ 2022 district opinion, 639 F. Supp. 3d at 954 (scraping can burden servers and inhibit a website’s performance); Ryanair DAC v. Booking Holdings Inc., 636 F. Supp. 3d 490, 503 (D. Del. 2022) (scraping can “greatly increase[] the quantifies of queries” on a website, “impair[] the . . . availability and/or usability” of a website, “and cause [a] website’s response times to deteriorate.”). Scraping can harm a website because a bot “can make many requests automatically and much more rapidly than any human could,” “request[ing] a huge amount of data from the target’s server.” See Compulife Software Inc. v. Newman, 959 F.3d 1288, 1299 (11th Cir. 2020). X Corp. has another case pending in this district, not before this Court, involving allegations of that kind of scraping. See X Corp. v. Bright Data Ltd., No. 23-cv-3698-WHA, First Amended Complaint (dkt. 36) (N.D. Cal. July 26, 2023) ¶ 1 (defendant “scrapes and sells millions of records from X Corp.’s X platform”), ¶ 98 (such actions “have diminished the server capacity that X Corp. can devote to its legitimate users, and thereby injured X Corp. by depriving it of the ability to use its personal property”), ¶ 102 (alleging “damage in the form of impaired condition, quality, and value of its servers, technology infrastructure, services, and reputation”); see also X Corp. v. John Doe 1, No. DC-23-09157, Original Petition (Tex. Dist. Ct., Dallas Co., July 6, 2023) ¶¶ 21–22, 24 (alleging “widespread unlawful scraping,” “flooding Twitter’s sign-up page with automated requests,” “severely tax[ing] X Corp.’s servers,” “impair[ing] the user experience for millions of X Corp.” users).

This is not such a case. Here, CCDH is alleged to have used Twitter’s own search tool to collect 9,615 public tweets from ten Twitter users, see FAC ¶ 77; Toxic Twitter at 17, and then to have announced that it did so in a public report, see id. Assuming for present purposes that this conduct amounts to the “scraping” barred by the ToS,[23] the extent of CCDH’s scraping was not a mystery. As CCDH asked at the motion hearing, “What CCDH did and the specific tweets that it gathered, what tool it used, how it used that tool and what the results were are documented explicitly in its public report. So what is it that they’re investigating?” Tr. of 2/29/24 Hearing at 22: 3–7. Nor was this the kind of large-scale, commercial scraping—as in hiQ, as alleged in Bright Data—that could conceivably harm the X platform or overburden its servers. It is not plausible that this small-scale, non-commercial scraping would prompt X Corp. to divert “dozens, if not over a hundred personnel hours across disciplines,” see Tr. of 2/29/24 Hearing at 8:7–11, of resources toward the repair of X Corp.’s systems. Nor would such expenditures have been foreseeable to CCDH in 2019. In 2019, if CCDH had thought about the no-scraping provision in the ToS at all, it would have expected X Corp. to incur damages only in response to breaches of that provision that could actually harm the X platform. It would not have expected X Corp. to incur damages in connection with a technical breach of that provision that involved the use of Twitter’s search tool to look at ten users and 9,615 public tweets.

It is clear to the Court that if X Corp. was indeed motived to spend money in response to CCDH’s scraping in 2023, it was not because of the harm such scraping posed to the X platform, but because of the harm it posed to X Corp.’s image. CCDH’s data collection allowed it to claim knowledge of what was occurring on the X platform in a cumulative sense. According to CCDH, its scraping revealed that X Corp. was generating millions of dollars in advertising revenue from previously banned accounts. See FAC ¶ 49; Kaplan Decl. Ex. B. X Corp.’s desire to keep entities like CCDH from drawing such unflattering conclusions about the X platform is entirely reasonable from a business point of view. But that motivation would not have been foreseeable to CCDH in 2019. In 2019, CCDH would not have understood, and should not have understood, that Twitter would wish to prevent scraping in order to keep cumulative information about the platform from the public view.[24] CCDH would certainly not have known that Twitter would change course on some of its content-moderation policies, reinstating what CCDH later described as “tens of thousands of accounts, including neo-Nazis, white supremacists, misogynists and spreaders of dangerous conspiracy theories,” see Toxic Twitter at 3, and then desire to shield the financial ramifications of that policy shift from the public. That is the kind of theoretical, speculative harm that California contract law rejects. See Erlich, 21 Cal. 4th at 550.

Neither the “safety and security” rationale, nor the “harm to X Corp.” rationale for X Corp. spending time and money in response to CCDH’s scraping make any sense given the scraping alleged in this case. Such damages simply were not foreseeable when CCDH entered into the ToS in 2019. Accordingly, allowing X Corp. to amend in order to add the proposed new allegations about damages in connection with its response to CCDH’s scraping would be futile.

The Court notes, too, that X Corp.’s motivation in bringing this case is evident. X Corp. has brought this case in order to punish CCDH for CCDH publications that criticized X Corp.—and perhaps in order to dissuade others who might wish to engage in such criticism.[25] Although X Corp. accuses CCDH of trying “to censor viewpoints that CCDH disagrees with,” FAC ¶ 20, it is X Corp. that demands “at least tens of millions of dollars” in damages—presumably enough to torpedo the operations of a small nonprofit—because of the views expressed in the nonprofit’s publications, see, e.g., id. ¶¶ 1, 3–4, 5, 12, 17–24, 38, 41, 43–52, 54–60, 62, 65–70, 77–78, 84–85, 92, 96, 98; see also PPP Br. at 1 (“X Corp. . . . seeks to silence critique rather than to counter it.”). If CCDH’s publications were defamatory, that would be one thing, but X Corp. has carefully avoided saying that they are.

Given these circumstances, the Court is concerned that X Corp.’s desire to amend its breach of contract claim has a dilatory motive—forcing CCDH to spend more time and money defending itself before it can hope to get out from under this potentially ruinous litigation. See PPP Br. at 2 (“Without early dismissal, the free speech interests that the California legislature sought to protect will vanish in piles of discovery motions.”). As CCDH argued at the motion hearing, the anti-SLAPP “statute recognizes that very often the litigation itself is the punishment.” Tr. of 2/29/24 Hearing at 33:12–34:5. It would be wrong to allow X Corp. to amend again when the damages it now alleges, and the damages it would like to allege, are so problematic, and when X Corp.’s motivation is so clear.

Accordingly, the Court STRIKES the breach of contract claim and will not allow X Corp. to add the proposed new allegations as to that claim.

[20] At the motion hearing, CCDH responded, not unreasonably, that “[i]t’s not exactly a[n] earth-shattering rule of law that contract damages have to be foreseeable and comport with the Constitution” and that, in “two shots at this,” X Corp. had only pled publication damages. See id. at 20:18–23.

[21] While the Court recognizes that technically, who sees a post might “depend[] on the privacy settings,” see Tr. of 2/29/24 Hearing at 40:9–16, that is a red herring here. The whole reason that CCDH wrote about these ten users is that a large audience viewed their tweets. See Toxic Twitter at 3 (“The analysis shows that the ten accounts have already amassed 2.5 billion tweet impressions”). It is not a reasonable inference that these ten users had selected privacy settings that limited their posts to friends and family.

[22] At times, it seemed as if X Corp. was suggesting that whenever a user breaches any term within X Corp.’s nineteen-page ToS, X Corp. is entitled to spend money to prevent further breaches of that term, and then to be compensated for that expenditure. That is not the law. If it was, then a plaintiff could state a claim merely by alleging the existence of a contract, its performance, and the defendant’s breach; there would be no separate requirement of proving foreseeable damages. But there is such a requirement. See Oasis W. Realty, 250 P.3d at 1121; Erlich, 21 Cal. 4th at 550.

[23] CCDH argues that it does not. See MTD&S at 10–13.

[24] CCDH framed this point thusly at the motion hearing: “the notion that CCDH . . . should have foreseen Twitter’s costs in better hiding white supremacist content on the platform is utterly implausible as a matter of law.” Tr. of 2/29/24 Hearing at 36:8–11.

[25] “A recent survey of 167 academics and researchers found that over 100 studies about X Corp. have been diverted, stalled, or canceled, with over half of those interviewed citing a fear of being sued by X Corp. over their findings or data.” ACLU Br. at 14 (citing Sheila Dang, “Exclusive: Elon Musk’s X Restructuring Curtails Disinformation Research, Spurs Legal Fears,” Reuters (Nov. 6, 2023), https://www.reuters.com/technology/elon-musks-xrestructuring-curtailsdisinformation-research-spurs-legal-fears-2023-11-06).