Opinion
21-60285
10-25-2024
Tesla, Incorporated, Petitioner Cross-Respondent, v. National Labor Relations Board, Respondent Cross-Petitioner. International Union, United Automobile, Aerospace and Agricultural Implement Workers of America, AFL-CIO, Petitioner,
Petition for Review of an Order of the National Labor Relations Board Agency Nos. 32-CA-197020, 32-CA-197058, 32-CA-197091, 32-CA-197197, 32-CA-200530, 32-CA-208614, 32-CA-210879, 32-CA-220777
Before Elrod, Chief Judge, and Jones, Smith, Stewart, Dennis, Richman, Southwick, Haynes, Graves, Higginson,
Willett, Duncan, Engelhardt, Oldham, Wilson, Douglas, and Ramirez, Circuit Judges. [*]
JUDGE HAYNES concurs in the judgment only.
This case arises from a union organizing campaign at one of Tesla's electric vehicle factories. Three pro-union Tesla employees and the United Auto Workers ("UAW") alleged that Tesla engaged in unfair labor practices in violation of the National Labor Relations Act ("NLRA"). They filed charges with the National Labor Relations Board ("NLRB"), which found that a tweet by Tesla CEO Elon Musk was an unlawful threat and ordered it deleted. The Board likewise found that Tesla employee Richard Ortiz had been terminated in violation of the NLRA. Tesla petitioned this court for review of that order, and the NLRB cross-petitioned for enforcement. A panel of this court affirmed the Board's order and granted its petition for enforcement. Tesla, Inc. v. NLRB, 63 F.4th 981 (5th Cir.), reh'g en banc granted, opinion vacated, 73 F.4th 960 (5th Cir. 2023) (mem.). Tesla petitioned for rehearing en banc, which we granted. Tesla, 73 F.4th 960.
We VACATE the Board's order and REMAND for further proceedings.
I
We briefly recount (A) the history of Musk's tweets, then (B) the story of Ortiz's termination.
A
Musk posted a photograph on Twitter of a rocket belonging to one of his companies, Space Exploration Technologies Corp. A Twitter user named "dmatkins137"-who was not a Tesla employee-responded to that post with an article from a publication called "Reveal." The "Reveal article" claimed that Tesla maintained unsafe working conditions. Musk and "dmatkins137" then engaged in the following colloquy:
The website formerly known as "Twitter" is now "X." For clarity, however, we will refer to it as Twitter throughout this opinion because this case arose before the name change.
@elonmusk: Tesla factory literally has miles of painted yellow lines &tape. Report about forklifts not beeping is also bs. These are both demonstrably false, but were reported as "facts" by Reveal.
@dmatkins137: Yellow is fine, got it. How about unions?
@elonmusk: Nothing stopping Tesla team at our car plant from voting union. Could do so tmrw if they wanted. But why pay union dues &give up stock options for nothing? Our safety record is 2X better than when plant was UAW &everybody already gets healthcare.ROA.4536-37.
Two days later, another Twitter user named "ericbrownzzz"-also not a Tesla employee-replied to Musk's original tweet. "ericbrownzzz" accused Musk of threatening to take away employee benefits, and Musk again responded:
@ericbrownzzz: Hi Elon, why would they lose stock options? Are you threatening to take away benefits from unionized workers?
@elonmusk: No, UAW does that. They want divisiveness &enforcement of 2 class "lords &commoners" system. That sucks. U.S. fought War of Independence to get *rid* of a 2 classsystem! Managers &workers shd be equal w easy movement either way. Managing sucks btw. Hate doing it so much.ROA.4537. Musk further stated that "UAW does not have individual stock ownership as part of the compensation at any other company," and as such, Tesla employees would lose stock options if they unionized because of UAW's policy. ROA.4539.
UAW filed an unfair-labor-practice charge based on only Musk's original tweet, responding to "dmatkins137." The Union alleged that the tweet was a threat to rescind stock options if employees unionized, and therefore violated Section 8(a)(1) of the NLRA.
The Board agreed. The NLRB believed that Musk's tweet violated the standard set forth in NLRB v. Gissel Packing Co., 395 U.S. 575 (1969). And it ordered Musk to delete his tweet.
B
Around the same time, UAW campaigned to unionize Tesla's facility in Fremont, California. Tesla employees Richard Ortiz and Jose Moran were among the most active supporters of UAW's campaign. During that time, three Tesla employees testified before the California legislature against union-backed legislation.
Ortiz did not personally watch these proceedings but wanted to find out which of his coworkers testified against the legislation. So he asked Moran to watch the hearings and identify the employees who testified. Moran sent Ortiz the names of the employees and their photos from Tesla's internal Workday website. Ortiz then posted the employees' photos on a pro-union Facebook group, along with the following statement:
These guys been in Sacramento saying we are lying about how things are at Tesla Management has been taking them one of them sez [sic] he made $130000 last year .... How many of you make . . . overtime .... This just proves how much kissing ass and ratting on people get you at Tesla and the ones that do the real work get passed over ....ROA.6278.
One of the outed employees complained to Tesla Human Resources. The employee said Ortiz's post was harassment.
Ricky Gecewich opened a human resources investigation. Gecewich asked Ortiz if he had taken screenshots of the Workday photos to post them on Facebook. Ortiz denied doing so. When asked where he obtained the pictures, Ortiz said he "didn't know" and "wasn't sure." ROA.535, 537.
After investigating computer records, Tesla learned that Moran sent Ortiz the screenshots. In light of this information, Gecewich again asked Ortiz how he got the photographs. At this time, Ortiz admitted that he lied during his first conversation with Gecewich about how he obtained the pictures. Gecewich then recommended that Tesla terminate Ortiz for "knowingly" misleading the investigation and because he "lied about knowing the source of the Workday screenshots" and "admitted to not telling the truth during the investigation interviews." ROA.4508.
Tesla referred Gecewich's recommendation to an independent decisionmaker, Stephen Graminger. Before making any termination decision, Graminger asked how Tesla handled dishonesty during previous internal investigations. He decided to terminate Ortiz after determining that Tesla had previously terminated an employee for lying to the company.
The NLRB's General Counsel and UAW filed unfair labor practice charges against Tesla for terminating Ortiz. The Board determined that "[t]he credited evidence show[ed] that [Tesla] terminated Ortiz for lying during an investigation." ROA.6285. But it nevertheless found that the company violated the NLRA. The Board based that finding on Tesla's anti-union animus. The Board ordered that Tesla reinstate Ortiz with backpay.
* * *
Tesla petitioned this court for review of the Board's order, which addressed both Musk's tweet and Ortiz's termination. A panel of this court agreed with the Board. See Tesla, 63 F.4th at 996. Tesla petitioned for en banc rehearing, which we granted.
II
We (A) hold that the agency exceeded its authority by ordering Musk to delete his tweet. We then (B) remand to the agency to consider the fact that the actual decisionmaker in Ortiz's firing harbored no anti-union animus.
A
The NLRB erred in ordering the deletion of Musk's speech as a remedy for unfair labor practices. That alone is enough to vacate its order, so we do not reach the merits of whether the tweet constituted an NLRA violation.
Deleting the speech of private citizens on topics of public concern is not a remedy traditionally countenanced by American law. "[T]he remedy" for bad speech, after all, should be "more speech, not enforced silence." Whitney v. California, 274 U.S. 357, 377 (1927) (Brandeis, J., concurring); cf. Frisby v. Schultz, 487 U.S. 474, 485 (1988) (explaining that a speech restriction is narrowly tailored only "if it targets and eliminates no more than the exact source of the 'evil' it seeks to remedy." (quoting City Council of L.A. v. Taxpayers for Vincent, 466 U.S. 789, 808 (1984))).
By contrast, deletion is a remedy for communications that are, in the First Amendment's contemplation, not speech at all. For example, the government can order the destruction of obscene material, perjurious material, or other non-speech material. See Kingsley Books, Inc. v. Brown, 354 U.S. 436, 444 (1957) (upholding a "Penal Law [that] provides for destruction of obscene matter following conviction for its dissemination"); R.A.V. v. City of St. Paul, 505 U.S. 377, 383 (1992) (explaining that certain communications fall outside the scope of the First Amendment and are "not . . . speech at all" (citation omitted)); see also United States v. Alvarez, 567 U.S. 709, 717-18 (2012) (plurality) (listing categories).
These same rules apply to labor disputes. Consider Gissel, for example. That case involved pamphlets that the Board found unlawful under the NLRA. 395 U.S. at 587-89. Rather than order the offending pamphlets destroyed, the Board issued a bargaining order against the corporation. See id. at 591-92; see also 29 U.S.C. § 160(a). Such a sanction fell within the traditional remedial ambit of the First Amendment by punishing the speaker for wrongful speech, rather than destroying the communications.
In this case, we assume without deciding that the NLRA applies to speech on Twitter. And we further assume without deciding that Musk's tweet violated the NLRA. Even so, the Board's speech-deletion order cannot stand. We hold that Musk's tweets are constitutionally protected speech and do not fall into the categories of unprotected communication like obscenity and perjury. And the Board does not dispute the general rule that it (like every other part of the Government) is powerless to delete protected speech. Rather, the Board's only authority for its speech-deletion remedy is a previous NLRB opinion involving speech on Twitter, FDRLST Media, 370 N.L.R.B. No. 49, at 1 n.5 (2020), which was subsequently vacated by the Third Circuit. See FDRLST Media, LLC v. NLRB, 35 F.4th 108, 126 (3d Cir. 2022) (vacating the NLRB's finding that a Tweet was an unfair labor practice and holding that when "protecting employees' statutory labor rights, neither we, nor the Board, can violate an employer's right to free speech under the First Amendment"). We follow our sister circuit in vacating the Board's decision here.
The dissent accuses us of issuing a "logically incoherent en banc opinion given that finding a violation is a condition precedent to the issue of remedy." Post, at 12; see also id. at 20. True, we "expressly pretermit[] whether Musk's tweet was an unfair labor practice," id. at 12, because we hold that the tweet-deletion injunction cannot issue. Pretermit-ting the merits because the remedy cannot issue is far from logically incoherent. Federal courts make this move in many areas of law. See, e.g., Pearson v. Callahan, 555 U.S. 223, 242 (2009) (leaving "to the sound discretion of the lower courts to determine the order of" deciding (1) whether the government has violated the Fourth Amendment and (2) whether to deny suppression under the good-faith exception). And independent bars to injunctive relief are often a reason not to reach the merits of a case. See, e.g., Whole Woman's Health v. Jackson, 595 U.S. 30, 44-45 (2021) (pretermitting the merits of Texas S.B. 8's constitutionality where no Ex parte Young action was available against the Texas attorney general); Younger v. Harris, 401 U.S. 37, 41 (1971) (declining to rule on the "constitutionality of the state law" sought to be enforced by the district attorney because "of the national policy forbidding federal courts to stay or enjoin pending state court proceedings except under special circumstances"); Jones v. Hendrix, 599 U.S. 465, 471-80 (2023) (as in much of habeas law, declining to reach the merits of postconviction relief claim where the writ of habeas is barred under AEDPA); Mississippi v. Johnson, 71 U.S. (4 Wall.) 475, 497-501 (1867) (declining to address the constitutionality of the Reconstruction Acts since the injunction against the President could not issue).
Before the en banc court, the agency offers a new argument. That is something this court cannot permit. See Calcutt v. FDIC, 598 U.S. 623, 628-29 (2023) (per curiam); see also SEC v. Chenery Corp., 332 U.S. 194, 196 (1947) (holding that if the agency rests its decision on "grounds [that] are inadequate or improper, the court is powerless to affirm the administrative action by substituting what it considers to be a more adequate or proper basis"). Even if Chenery did not forbid us from accepting the NLRB's argument, it is still unavailing. The agency now justifies the speech deletion as a "customary" exercise of the Board's power "to expunge a violation." NLRB EB Br. at 42. To support that "customary" power, the Board points to a case where it ordered an employer to remove from its files references to an unlawful discharge. Id. at 42 n.20 (citing Cordua Rests., Inc., 368 N.L.R.B. No. 43, at *7 (2019)). Whatever power the Board might or might not have to order such remedies involving a company's non-public, internal files, it says nothing about the Board's authority over public speech on topics of public concern. See Snyder v. Phelps, 562 U.S. 443, 452 (2011) ("[S]peech on public issues occupies the highest rung of the hierarchy of First Amendment values, and is entitled to special protection." (quoting Connick v. Myers, 461 U.S. 138, 145 (1983))).
B
Finally, we consider the Board's decision to order Tesla to reinstate Ortiz with backpay.
We review the Board's findings for substantial evidence. Dish Network Corp. v. NLRB, 953 F.3d 370, 376 (5th Cir. 2020). Substantial evidence is a term of art defined both in the NLRA and through years of Supreme Court precedent. Id.; see also 29 U.S.C. § 160(e)-(f) (establishing that substantial evidence means a review of "the record considered as a whole"). The Supreme Court has explained: "The substantiality of evidence must take into account whatever in the record fairly detracts from its weight." Universal Camera Corp. v. NLRB, 340 U.S. 474, 488 (1951). And it has explained that "the requirement for canvassing 'the whole record' in order to ascertain substantiality" means that Congress has
made it clear that a reviewing court is not barred from setting aside a Board decision when it cannot conscientiously find that the evidence supporting that decision is substantial, when viewed in the light that the record in its entirety furnishes, including the body of evidence opposed to the Board's view.Id. What that means for our court is that the evidence from the Board "must be substantial, not speculative, nor derived from inferences upon inferences." Brown &Root, Inc. v. NLRB, 333 F.3d 628, 639 (5th Cir. 2003). So "to survive substantial evidence review, then, the Board has to consider 'contradictory evidence or evidence from which conflicting inferences could be drawn.'" Dish Network Corp., 953 F.3d at 377 (quoting Universal Camera, 340 U.S. at 487).
Here, the NLRB failed to consider the contradictory fact that the decisionmaker who authorized Oritz's firing, Graminger, had no anti-union animus. The sworn testimony before the Board on the subject went as follows:
Q: Was Mr. Ortiz's union activity-did that play any role whatsoever in the decision you made to terminate Mr. Ortiz?
A: No, not at all.ROA.1309-10. In fact, the record shows that Graminger was a member of a different union and harbored pro-union sentiments. He testified that he was interested in "[t]he efforts to make UAW and . . . because I'm still a member of the union in Germany and I still pay my membership fees-I'm interested in it." ROA.1331. The Board did not consider these facts, which detract from its conclusion, and hence its order must be vacated. See Dish Network Corp., 953 F.3d at 377.
The NLRB and UAW resist this conclusion by arguing that Tesla cannot rely on Graminger as a neutral decisionmaker because he received some incorrect information from Gecewich. It is true that an employer cannot avoid liability by pointing to a neutral decisionmaker if another decisionmaker "had a significant role in the discharge and [used] anti-union discrimination [to] infect[] what might otherwise have been an innocent sterile act." NLRB v. Neuhoff Bros., Packers, Inc., 375 F.2d 372, 375 (5th Cir. 1967).
Here, however, Gecewich's misstatements were minor and did not play a "significant role" in Ortiz's termination. For example, Gecewich confused which employee complained about Workday misuse, the private nature of the Facebook group where Ortiz posted the Workday pictures, and the nature of the employees' legislative testimony. None of the alleged errors relate to the essential facts that Ortiz lied, and that Tesla had previously terminated an employee for dishonesty during company investigations. That is a far cry from Neuhoff, where an anti-union supervisor threatened to "fire anyone [who] sign[ed] a union card," personally threatened to fire a particular employee who signed a union card, and then "categorically" told the ultimate decisionmaker to discharge that employee. Id. at 374-76.
On remand, the Board is free to reconsider the record and make any decision supported by substantial evidence. It bears emphasis, however, that the Board bears the burden of showing "that the employer acted out of antiunion animus." Valmont Indus., Inc. v. NLRB, 244 F.3d 454, 464 n.2 (5th Cir. 2001). This means the Board "must do more than simply support an inference that protected conduct is a motivating factor in the employer's decision." Id.
* * *
For the foregoing reasons, the NLRB's order is VACATED, Tesla's petition for review is GRANTED, the Board's petition for enforcement is DENIED, and the case is REMANDED to the NLRB. The Board's crosspetition for enforcement is DENIED AS MOOT.
JAMES L. DENNIS, Circuit Judge, joined by STEWART, RICHMAN, SOUTHWICK, GRAVES, HIGGINSON, DOUGLAS, and RAMIREZ, Circuit Judges, dissenting:
In a short opinion that is light on law and facts, the en banc plurality only reaches two of the four issues on appeal and punts on the rest. It holds that the National Labor Relations Board (NLRB)'s order directing Elon Musk to delete a coercive tweet (a threat to take stock options away from Tesla employees if they voted to become unionized) violates the First Amendment, against the weight of four Supreme Court authorities. See NLRB v. Va. Elec. &Power Co., 314 U.S. 469 (1941); Thomas v. Collins, 323 U.S. 516 (1945); NLRB v. Gissel Packing Co., 395 U.S. 575 (1969); Chamber of Com. v. Brown, 554 U.S. 60 (2008). It also holds that pro-union Tesla employee Richard Ortiz may not have been unlawfully terminated because Tesla's ultimate decision-maker, a supervisor-employee, testified that he harbored no anti-union animus-testimony the NLRB considered and rejected. The plurality says zip about whether the NLRB is entitled to enforcement of the seven uncontested Tesla labor violations found by the NLRB. And the plurality expressly pretermits whether Musk's tweet was an unfair labor practice (ULP), resulting in a logically incoherent en banc opinion given that finding a violation is a condition precedent to the issue of remedy in this case. See NLRB v. Riley-Beaird, Inc., 681 F.2d 1083, 1086 (5th Cir. 1982) (holding that whether speech is protected by the First Amendment is an identical inquiry to whether it constituted a threat because the First Amendment's protections end where threats in violation of the NLRA begin).
The plurality's approach is inconsistent with established First Amendment principles and with this court's role as a court of review. I would grant the NLRB's petition to enforce its findings of the seven ULPs in which Tesla has acquiesced; find that its factual findings with respect to Musk's tweet and Ortiz's termination are supported by substantial evidence; and find that it did not exceed or abuse its broad remedial authority-to take action to "effectuate the policies of" the National Labor Relations Act (NLRA), see 29 U.S.C. § 160(c)-in ordering Musk to delete the threatening tweet. I therefore respectfully dissent.
I
The plurality omits several material details from its opinion. In summer 2016, Tesla employee Jose Moran contacted the United Auto Workers union (UAW) about unionizing Tesla employees at the technology and design corporation's car manufacturing facility in Fremont, California. As part of the unionization effort, the UAW created a Voluntary Organizing Committee (VOC) of employees acting as union organizers. Along with Moran and others, Richard Ortiz, Jonathan Galescu, and Michael Sanchez were VOC members. As retribution for efforts to organize, Tesla repeatedly violated the NLRA by coercing and discriminating against supporters of the union. Specifically, as affirmed by the NLRB, an Administrative Law Judge (ALJ) found that Tesla: (1) interfered with multiple employees' leafletting activities in February and May 2017; (2) prohibited employees from distributing union materials without approval and threatened them with discharge on March 23, 2017; (3) threatened on March 30, 2018, that selecting the UAW as a bargaining representative would be futile; (4) prohibited employees from communicating with the media about their employment by requiring employees to sign a Confidentiality Agreement containing a provision that stated "it is never OK to communicate with the media" about Tesla; (5) coercively interrogated employees about their protected union activities; (6) promulgated a rule restricting employees' use of Tesla's "Workday" program in response to Ortiz and Moran's protected union activities; (7) disciplined Moran for his union activity; (8) unlawfully terminated Ortiz for engaging in protected union activity; and (9) threatened to take stock options away from Tesla employees if they voted to became unionized by means of a tweet by Elon Musk on Twitter (now known as "X"). This en banc appeal centers around only the two ULPs that Tesla contests on appeal: the threatening tweet by Musk; and the termination of Tesla employee and union activist Richard Ortiz.
"Workday is a third-party [human resources] software program that [Tesla] uses to electronically store and access employees' personnel files. Employees can access Workday to, among other things, view and electronically sign documents."
A
Tesla's CEO, agent, and supervisor Elon Musk maintains the Twitter handle ""elonmusk" as his personal account and uses it to tweet about Tesla's business decisions and plans, finances, production goals, personnel matters, and breaking news. On May 20, 2018, during the organizational campaign, Musk tweeted:
"[T]he social media platform Twitter allows its users to publish short messages, photographs, videos, and hyperlinks (all called 'tweets') to the general public. Other users may respond to or republish those tweets and engage in virtual dialogues with other users on the platform." Campbell v. Reisch, 986 F.3d 822, 823 (8th Cir. 2021).
Nothing stopping Tesla team at our car plant from voting union. Could do so tmrw if they wanted. But why pay union dues &give up stock options for nothing? Our safety record is 2X better than when plant was UAW &everybody already gets healthcare.Musk's tweet was in response to another user, who asked Musk, in part, "How about unions?". Over the next few days, other users and Musk interacted on the same "thread" of tweets, as well as on an additional thread, in what Musk asserts was his effort to clarify his earlier tweet. Musk's attempted clarification came days after the original May 20 tweet.
When one looks at a tweet, "[a] comment thread appears below the original tweet and includes both the first-level replies (replies to the original tweet) and second-level replies (replies to the first-level replies)." Knight First Amend. Inst. Columbia Univ. v. Trump, 928 F.3d 226, 230 (2d Cir. 2019), vacated as moot sub nom. Biden v. Knight First Amend. Inst. Columbia Univ., 141 S.Ct. 1220 (2021). Twitter threads thus "reflect multiple overlapping conversations among and across groups of users and are a large part of what makes Twitter a social media platform." Id. (citation and quotation marks omitted). Twitter threads become a permanent record of what the users said on that occasion.
On May 23, 2018, the UAW filed a ULP charge based on the May 20 tweet, alleging that Musk's tweet was a threat to rescind employees' stock options if the employees voted to unionize, a violation of Section 8(a)(1) of the NLRA.
B
Around the same time as Musk's tweet, at Tesla's behest, three employees opposed to unionization, including Travis Pratt, testified against proposed legislation supported by the UAW during a California legislature public hearing. Ortiz did not attend the hearing, but it was recorded. A political organizer on behalf of the UAW sent Ortiz a link to the video recording. Ortiz had difficulty accessing the recording, so he sent the link to Moran and asked if Moran could open it. Ortiz also asked Moran if he knew who the three employees were. Using his personal phone, Moran watched the video, noted the names of the employees, and used Tesla's Workday program to search for the employees' names to verify that they were in fact Tesla employees. Moran took screenshots of the Workday profiles of the three employees and sent them to Ortiz. The Workday profiles included a photo of each employee. Tesla had no policy prohibiting such use of Workday or otherwise restricting employee access to the program at that time.
At the time of this case, Tesla had no rule prohibiting Moran from using Workday to confirm the identity of the three pro-company employees who appeared in the legislative hearing against the UAW-sponsored legislation. Nor was there any company rule against Ortiz's use of the information gathered by Moran to inform pro-union employees of Tesla's agents-employees' anti-union legislative lobbying activity.
Ortiz posted two of the screenshots, including a screenshot of Pratt's profile, to a private "Tesla Employees for UAW Representation" Facebook page and included his comment that the pictured employees were "in Sacramento saying we are lying about how things are at Tesla." Ortiz noted that Pratt testified at the public hearing that his salary was $130,000, and Ortiz commented, "[t]his just proves how much kissing ass and ratting on people get you at Tesla and the ones that do the real work get passed over." Though the Facebook group was private, and Pratt was not a member, someone sent him the post. Pratt then sent Ortiz a message, objecting to the "name calling," after which Ortiz quickly removed the post from Facebook.
Employees interested in unionizing voted on a campaign slogan, "Driving a Fair Future at Tesla," and the UAW created a public website and public Facebook group, "A Fair Future at Tesla," in support of the campaign. Moran also created a private Facebook group called "Tesla Employees for UAW Representation." While the public group could be joined and viewed by anyone on Facebook, access to the private group was restricted and required approval from Moran or Ortiz.
Pratt also texted a screenshot of the post to Josh Hedges, a Senior Human Resources Director for Production and Supply Chain at Tesla, with the caption, "[l]ooks like we got under some people's skin," followed by a smiley face, referring to the testimony. Hedges asked whether the post was on Facebook, and Pratt responded, "Yea lol [laugh out loud] I'm pretty sure it's on their fair future at Tesla thing." Pratt allegedly told Hedges by phone that he felt harassed and targeted because of the Facebook post. After speaking with Pratt, Hedges submitted a complaint to Tesla's employeerelations team and contacted Tesla investigator Ricky Gecewich about the situation. A few days later, Gecewich interviewed Pratt, who repeated his story.
Pratt did not testify before the ALJ. Ortiz, Hedges, Moran, and Gecewich did. The ALJ found Moran credible, found Ortiz credible in part, and found Hedges and Gecewich not credible, assigning reasons for each credibility finding.
Gecewich also interviewed Ortiz, who said he apologized and removed the post after Pratt contacted him. Gecewich then asked where the pictures came from and Ortiz said he could not remember, which he later admitted was untrue. Gecewich obtained logs of who had viewed Pratt's Workday profile and identified Moran. Gecewich met with Moran, who said he accessed Pratt's Workday profile to confirm that he was a Tesla employee after seeing his testimony at the legislature. Moran told Gecewich that he needed to identify anti-union employees as part of the unionization campaign. He also told Gecewich he sent the screenshots to Ortiz. Gecewich met with Ortiz again, and Ortiz admitted that he had feigned lack of memory to protect Moran's identity.
Gecewich penned a report recommending that Ortiz be fired for "admittedly lying," and that Moran be disciplined "for accessing Workday for non-business related purposes." The report said Moran claimed he was asked by a UAW representative to verify whether Pratt and others were Tesla employees. The report also said that Moran had admitted to using Workday for other personal purposes in the past-for example, to compare his title to other employees. Hedges agreed with Gecewich's recommendation that Ortiz be fired and chose Stephan Graminger, Tesla's Director of Body Manufacturing, to be the ultimate decisionmaker as to Ortiz's fate.
Gecewich met with Graminger, along with Ortiz's direct manager, Ron Martinez, and another human resources official. Gecewich told Graminger that Ortiz had leaked personal information and lied during the investigation. Graminger did not make an immediate decision, but first asked his superior, Peter Hochholdinger, whether similar cases involving lying during an investigation had resulted in termination and was told that they had resulted in termination. After receiving this information, Graminger approved the decision to fire Ortiz. Ortiz's employment at Tesla was terminated on October 18, 2017. Ortiz's termination was not based on any specific human resources rule or policy.
The ALJ found Graminger to be a "more credible witness" than Hedges and Gecewich, but still found his testimony confusing and contradictory.
C
The UAW and three pro-union Tesla employees filed multiple charges with the NLRB alleging ULPs against Tesla. An ALJ found that Tesla had committed nine violations, and the NLRB issued an order affirming the ALJ. In our court, Tesla filed a petition for review of two of the nine ULPs and the NLRB filed a cross-application to enforce its order. A panel of our court issued a per curiam opinion denying Tesla's petition and granting the NLRB's cross-application. See Tesla, Inc. v. NLRB, 63 F.4th 981 (5th Cir. 2023), reh'g en banc granted, opinion vacated, 73 F.4th 960 (5th Cir. 2023). Our en banc court granted rehearing.
On en banc rehearing, Tesla continues to only challenge two of the nine labor violations found by the NLRB and the NLRB's tweet-deletion remedy. First, Tesla challenges the NLRB's finding that Musk committed a labor violation by posting an unlawful threat on Twitter. Second, Tesla contends that part of the NLRB's choice of remedy for this violation-an order for Tesla to direct Musk to delete the tweet-was "improper." And third, Tesla objects to the NLRB's finding that employee Richard Ortiz was unlawfully terminated.
II
Most puzzling among the plurality's errors is its silent declination of the NLRB's request for summary enforcement of the seven labor violations that Tesla chose not to contest on appeal. The NLRB is clearly entitled to summary enforcement of its order pertaining to the following Tesla violations: (1) interfering with employee leafletting; (2) prohibiting employees from distributing union materials without approval/threatening them with discharge; (3) threatening that selecting the UAW would be futile; (4) prohibiting employees from communicating with the media about their employment; (5) interrogating certain employees about union activity; (6) promulgating a rule restricting Workday use in response to Ortiz and Moran's union activity; and (7) disciplining Moran for his union activity. Tesla does not challenge these findings, and "[f]indings of the Board that the employer does not challenge are waived on review, entitling the Board to summary enforcement." Cordua Rests., Inc. v. NLRB, 985 F.3d 415, 422 (5th Cir. 2021) (citation omitted). The plurality's refusal to grant the NLRB's request for enforcement of these ULP findings is devoid of any support or explanation in its opinion or the record.
III
The ALJ found that Musk's tweet violated Section 8(a)(1) of the NLRA because it could be reasonably understood by employees as a threat to unilaterally rescind stock options if employees unionized, rather than as a carefully phrased prediction, based on objective fact, of the likely consequences of unionization beyond Tesla's control. The NLRB affirmed the ALJ's findings and ordered Tesla to cease and desist from its conduct and to direct Musk to delete the tweet from his Twitter account. Tesla argues that the tweet, especially when viewed in context, was not a threat and was instead protected by Section 8(c) of the NLRA. The en banc plurality pretermits this issue for no stated reason, even though it vacates the corresponding remedy on First Amendment grounds. Ante, at 7-8 (plurality opinion). I would reach the merits of the ULP issue because finding a labor violation is a condition precedent to addressing the issue of the tweet deletion remedy. See Riley-Beaird, Inc., 681 F.2d at 1086.
As explained, the tweet at issue stated "[n]othing stopping Tesla team at our car plant from voting union. Could do so tmrw if they wanted. But why pay union dues & give up stock options for nothing? Our safety record is 2X better than when plant was UAW & everybody already gets healthcare."
Since the NLRA's inception, Section 7 has afforded employees "the right to self-organization, to form, join, or assist labor organizations, to bargain collectively through representatives of their own choosing, and to engage in other concerted activities for the purpose of collective bargaining or other mutual aid or protection"-and "the right to refrain from any or all of such activities." See 29 U.S.C. § 157. To make this guarantee effective, Section 8(a)(1) makes it a ULP for an employer "to interfere with, restrain, or coerce employees in the exercise of the rights guaranteed in" Section 7. See 29 U.S.C. § 158(a)(1).
"An unlawful threat is established under § 8(a)(1), if under the totality of the circumstances, an employee could reasonably conclude that the employer is threatening economic reprisals if the employee supports the union." NLRB v. Delta Gas, Inc., 840 F.2d 309, 311 (5th Cir. 1988) (citation omitted). "The test for determining 'whether an employer has violated § 8(a)(1) is whether the employer's questions, threats or statements tend to be coercive, not whether the employees are in fact coerced[.]'" Brown &Root, Inc. v. NLRB, 333 F.3d 628, 634 (5th Cir. 2003) (quoting NLRB v. PNEU Elec., Inc., 309 F.3d 843, 850 (5th Cir. 2002)). Whether an employer is making an unlawful threat is measured objectively, from the perspective of an employee, and is not contingent on "either the motivation behind the remark or its actual effect." Miller Elec. Pump &Plumbing, 334 N.L.R.B. 824, 824 (2001); Brown &Root, 333 F.3d at 634.
Section 8(c) of the NLRA cabins Section 8(a)(1) by stating that "[t]he expressing of any views, argument, or opinion, or the dissemination thereof . . . shall not constitute or be evidence of an unfair labor practice under any of the provisions of this subchapter, if such expression contains no threat of reprisal or force or promise of benefit." See 29 U.S.C. § 158(c). To fall within Section 8(a)(1)'s protection then, an employer's prediction of the effects of unionization "must be carefully phrased on the basis of objective fact to convey an employer's belief as to demonstrably probable consequences beyond his control." Gissel, 395 U.S. at 618. If the employer's statement instead carries "any implication that an employer may or may not take action solely on his own initiative" in response to unionization, then it is a "threat of retaliation." Id. (emphasis added). Thus, a statement implying that unionization will result in the loss of benefits, without some explanation or reference to the collective-bargaining process, economic necessity, or other objective facts, is a coercive threat, but is not a threat if made in the context, for example, of explaining that existing benefits may be traded away during the bargaining process. UNF W., Inc. v. NLRB, 844 F.3d 451, 458 (5th Cir. 2016). Put simply, "an employer is free only to tell 'what he reasonably believes will be the likely economic consequences of unionization that are outside his control,' and not 'threats of economic reprisal to be taken solely on his own volition.'" Gissel, 395 U.S. at 619 (quoting NLRB v. River Togs, Inc., 382 F.2d 198, 202 (2d Cir. 1967)). In this way, Section 8(c) of the NLRA balances "an employer's free speech right to communicate his views to his employees" against an employee's Section 7 right to associate freely and to be free of coercion. Id. at 617.
The NLRB is an expert in labor law, so our review of its orders is supposed to be "limited and deferential." In-N-Out Burger, Inc. v. NLRB, 894 F.3d 707, 714 (5th Cir. 2018). The NLRB's factual findings are "conclusive" so long as they are "supported by substantial evidence on the record considered as a whole." 29 U.S.C. § 160(e). "Substantial evidence is that which is relevant and sufficient for a reasonable mind to accept as adequate to support a conclusion. It is more than a mere scintilla, and less than a preponderance." IBEW, Loc. Unions 605 &985 v. NLRB, 973 F.3d 451, 457 (5th Cir. 2020) (citation omitted). "[T]he ALJ's decision must be upheld if a reasonable person could have found what the ALJ found, even if" we would have reached a different conclusion had we heard the case in the first instance. Standard Fittings Co. v. NLRB, 845 F.2d 1311, 1314 (5th Cir. 1988). "In determining whether the Board's factual findings are supported by the record, we do not make credibility determinations or reweigh the evidence." NLRB v. Allied Aviation Fueling, 490 F.3d 374, 378 (5th Cir. 2007); accord Ill. Cent. R.R. Co. v. Norfolk &W. Ry. Co., 385 U.S. 57, 69 (1966) ("It is not for the court [on substantial evidence review] to strike down conclusions that are reasonably drawn from the evidence and findings in the case."). "Only in the most rare and unusual cases will an appellate court conclude that a finding of fact made by the [NLRB] is not supported by substantial evidence." Flex Frac Logistics, L.L.C. v. NLRB, 746 F.3d 205, 208 (5th Cir. 2014) (quoting Merchs. Truck Line, Inc. v. NLRB, 577 F.2d 1011, 1014 n.3 (5th Cir. 1978)). The NLRB's legal conclusions are reviewed de novo. Id. at 207.
Throughout this dissent, I occasionally focus on the ALJ's decision as opposed to the NLRB's decision because the ALJ engaged in the initial fact-finding with which the NLRB largely agreed. See, e.g., NLRB v. Gulf States United Tel. Co., 694 F.2d 92, 95 (5th Cir. 1982) (focusing on the findings of the ALJ, which were later adopted by the NLRB).
* * *
At the outset, Tesla argues that Musk's tweet was not made in the labor relations context because it was posted on Twitter, a "natural forum for public debate." Given that forum, Tesla argues, the NLRA's proscriptions cannot reach Musk's speech. For support, Tesla relies on the Third Circuit's decision in FDRLST Media, LLC v. NLRB, 35 F.4th 108 (3d Cir. 2022).
A close reading of FDRLST Media, LLC reveals that the Third Circuit's opinion is inapposite to Tesla's argument, but instead supports my conclusion that Musk's tweet was made in the labor relations context. In FDRLST Media, LLC, the Third Circuit found that a satirical, nonsensical tweet (about sending writers and editors to the salt mine) by the executive officer of The Federalist was not a threat in a case where there was not "even a single example of labor-management tension." 35 F.4th at 123-24. In doing so, the Third Circuit highlighted that "[t]he record does not show that [the executive officer] ever used []his account to communicate with employees or that employees were required to follow it." Id. at 124. In the present case, by contrast, referencing the "Tesla team," Musk issued a public message on a public platform that reached over twenty-two million individuals-some of whom are employees of Tesla-advocating against unionization at Tesla during an ongoing and heated labor dispute. The parties stipulated that Twitter, and the use of tweets, is a commonly accepted form in which some companies announce news in lieu of, or in addition to, press releases. Tesla specifically stipulated that Musk uses his personal Twitter account to communicate Tesla's business decisions and plans, finances, production goals, personnel matters, and breaking news. Indeed, Tesla's head of human resources credibly testified that "she understood Musk to tweet on behalf of Tesla." And we know that at least one Tesla employee saw the tweet. I would find that Musk's dissemination on Twitter is akin to a company official issuing a press release to the public where anyone including employees may read the statement. See Vemco, Inc., 304 N.L.R.B. 911, 925, enforced in part by NLRB v. Vemco, Inc., 989 F.2d 1468 (6th Cir. 1993) (press release broadcasted to the public sufficiently communicated same to the employees). Just as there, the undisputed facts of this case demonstrate that Musk's tweet occurred in the labor relations context.
In that case, on the same day that unionized employees of a competitor magazine business walked off the job, the executive officer of The Federalist tweeted, "FYI @fdrlst first one of you tries to unionize I swear I'll send you back to the salt mine." FDRLST Media, LLC, 35 F.4th at 113. The Third Circuit found "[t]he tweet's suggestion that . . . [writers and editors] might be sent 'back' to work in a 'salt mine'" was "farcical." Id. at 123. "The image evoked-that of writers tapping away on laptops in dimly-lit mineshafts alongside salt deposits and workers swinging pickaxes-is as bizarre as it is comical." Id.
Of course, those facts are wholly distinguishable from those present in this case. Here, there was no element of humor in Musk's tweet and the instant case arose from a "tense union campaign" in the context of "Tesla's history of labor violations." Tesla, Inc., 63 F.4th at 986, 993.
Beyond that, Tesla's argument that Musk's May 20, 2018, tweet "was not threatening on its face" is unavailing because implied threats are threats under the NLRA. See, e.g., Tellepsen Pipeline Servs. Co. v. NLRB, 320 F.3d 554, 564 (5th Cir. 2003) ("[Employer's] statements . . . constitute[d] implied threats of reprisal for union activities in violation of section 8(a)(1)."); J.L.M. Inc., 312 N.L.R.B. 304, 305 (1993) (implied threat contained in an employer's posting violated the NLRA). Tesla claims that the tweet was not a threat because it started out by saying that there was "[n]othing stopping" employees from unionizing and it is a strain to characterize "give up stock options for nothing" as a threat, because, unlike the threat of plant closure, compensation is not within the employer's unilateral control once employees unionize and the parties engage in collective bargaining. However, because stock options are part of Tesla's employees' compensation, and because nothing in the tweet suggested that Tesla would be forced to end stock options on account of unionization, or that the UAW would be the cause of giving up stock options, substantial evidence supported the NLRB's conclusion that the tweet was an implied threat to end stock options as retaliation for unionization. The record, in fact, bears out that only Tesla has the power to unilaterally revoke stock option benefits and the UAW has no policy precluding stock option benefits for its members. See Gissel, 395 U.S. at 618 (requiring "demonstrably probable consequences beyond [the employer's] control" that rest on "objective fact"). Moreover, the statement in Musk's tweet is materially similar to other statements that the NLRB and our court have found to be threats. See, e.g., NLRB v. Bama Co., 353 F.2d 320, 323 (5th Cir. 1965) (involving a supervisor threat that "unionization would probably result" in lost benefits); Hendrix Mfg. Co. v. NLRB, 321 F.2d 100, 104 (5th Cir. 1963) ("[I]f the Union came in, the current profit sharing plan would be discontinued."); Intermedics, Inc., 262 N.L.R.B. 1407, 1411 (1982), enforced by NLRB v. Intermedics, Inc., 715 F.2d 1022 (5th Cir. 1983) ("[I]f the Company were to go union the employees would lose all their benefits.").
Tesla's next argument, that the tweet was not a threat because it was grounded in the objective fact that UAW-represented employees at other companies supposedly do not have stock options, misapplies the relevant law. The legal test to determine what constitutes a threat under the NLRA is whether the tweet would have been reasonably understood by employees as a threat and therefore whether it tended to be coercive, not whether employees would have been able to independently verify that the tweet was based in objective fact. Brown &Root, 333 F.3d at 634. The tweet itself did not include any objective facts that would lead a reasonable employee to conclude that the UAW, rather than Tesla, would be the cause of employees giving up stock options. UNF W., Inc., 844 F.3d at 458 ("[I]f the statement in its context fails to include any reference to the collective-bargaining process or to any economic necessities or other objective facts as a basis for its prediction . . . then it is impermissible, because it implies that an employer may act on its own initiative, unilaterally, and for its own reasons." (internal quotation omitted)).
And again, the record before us establishes that the UAW has no policy prohibiting stock option benefits for union members.
Furthermore, the remainder of the Twitter thread, a related thread, and a later press release do nothing to stop employees from reasonably interpreting Musk's May 20, 2018, tweet as a threat. Tesla postulates that the ALJ erred by not analyzing the tweet in the context of later tweets and communications, demonstrating that Musk believed the UAW would take away stock options and clarifying that the original tweet was not a threat. Tesla is incorrect for two reasons. First, Musk's May 22 and May 23 tweets and publications were not "contemporaneous" with his May 20 tweet, so they cannot change whether the original tweet was objectively an implied threat. Id. ("[C]ontemporaneous or earlier contextual factors can influence a statement's reasonable import for the listener at the time that the statement was uttered." (citation omitted)). Second, although "additional comments can be made to clarify, expand, or otherwise alter the context and reasonable import" of the original tweet, id., the parties stipulated that "[i]t is not possible to know or determine if every individual that viewed the tweets by Elon Musk [on] May 20 also viewed the tweets by Elon Musk [from] May 22 and 23." Combined with Tesla's history of labor violations, this supports the NLRB's finding that employees would understand Musk's tweet as a threat to retaliate by rescinding stock options; especially when considered in the context of "the economic dependence of [Tesla] employees on their employer[]." Gissel, 395 U.S. at 617.
On May 22, 2018, two days after the initial tweet, Musk was asked by another Twitter user, in a reply to his initial May 20 tweet, "[a]re you threatening to take away benefits from unionized workers?", to which he responded "No, UAW does that." The next day, May 23, writing on a different Twitter thread, Musk tweeted "UAW does not have individual stock ownership as part of the compensation at any other company." The users that Musk responded to were not Tesla employees.
Lastly, Tesla asks us to depart from our precedent and the precedent of every other circuit by arguing that the absence of subjective employee coercion evidence undercuts the NLRB's finding that Musk's tweet was coercive. But the test for determining "whether an employer has violated Section 8(a)(1) is whether the employer's questions, threats or statements tend to be coercive, not whether the employees are in fact coerced." Brown &Root, 333 F.3d at 634; see also, e.g., Allegheny Ludlum Corp. v. NLRB, 301 F.3d 167, 176 (3d Cir. 2002) ("[T]he test is an objective test in which the employer's intent is irrelevant and the proper inquiry is the impression of a reasonable employee."); Teamsters Loc. Union No. 171 v. NLRB, 863 F.2d 946, 954 (D.C. Cir. 1988); NLRB v. Marine Optical, Inc., 671 F.2d 11, 18 (1st Cir. 1982); HealthBridge Mgmt., LLC v. NLRB, 902 F.3d 37, 46 (2d Cir. 2018); Alton H. Piester, LLC v. NLRB, 591 F.3d 332, 336 (4th Cir. 2010); Torbitt &Castleman, Inc. v. NLRB, 123 F.3d 899, 906 (6th Cir. 1997); C&W Super Mkts., Inc. v. NLRB, 581 F.2d 618, 624 n.5 (7th Cir. 1978); Russell Stover Candies, Inc. v. NLRB, 551 F.2d 204, 207-08 (8th Cir. 1977); Lippincott Indus., Inc. v. NLRB, 661 F.2d 112, 114 (9th Cir. 1981); Lear Siegler Inc. v. NLRB, 890 F.2d 1573, 1580 (10th Cir. 1989); Mead Corp. v. NLRB, 697 F.2d 1013, 1025 (11th Cir. 1983). Tesla accurately recounts that in FDRLST Media, LLC, 35 F.4th at 125, the Third Circuit stated, "subjective responses can be relevant" in a case where "a third party with no connection to the employer or the employees-and who lack[ed] knowledge of the relevant context-file[d] a charge against an employer with no history of labor problems." Ours, however, is not that case. See supra p. 12 n.11. In cases like this one, where employees participating in an active organizing campaign allege coercion based on a non-joking statement from an employer with a history of ULPs, courts have never required evidence of subjective employee coercion. I would decline Tesla's invitation to be the first.
At bottom, this is far from "the most rare and unusual case[]" where we will "conclude that a finding of fact made by the [NLRB] is not supported by substantial evidence." Flex Frac Logistics, L.L.C., 746 F.3d at 208 (citing Merchs. Truck Line, Inc., 577 F.2d at 1014 n.3). When an employer threatens to do something that is only within its power, employees will take their employer seriously because their livelihoods are on the line. See Gissel, 395 U.S. at 617 ("[T]he economic dependence of the employees on their employers, and the necessary tendency of the former, because of that relationship, to pick up intended implications of the latter that might be more readily dismissed by a more disinterested ear."). That rings true, especially when, as here, the employer makes that statement during a heated union campaign and when the employer has a history of committing labor violations (Tesla acquiesces in seven other labor violations in this case alone). I would conclude that substantial evidence supported the NLRB's finding that Musk's May 20, 2018, tweet was a threat to unilaterally rescind stock options if employees unionized.
IV
Despite "assum[ing] without deciding that Musk's tweet" was a ULP, the plurality proceeds to vacate the NLRB's order directing Musk to delete the offending tweet because the remedy is novel and violative of the First Amendment. Ante, at 6-8. Respectfully, both arguments are misplaced.
First, in Section 10(c) of the NLRA, Congress gave the NLRB broad remedial authority to order violators "to cease and desist" from their ULPs and "take such affirmative action" as will "effectuate the policies of" the NLRA. See 29 U.S.C. § 160(c). The Supreme Court and our court apply a very liberal standard when reviewing challenges to the NLRB's choice of remedy. The NLRB's choice of remedy "must be upheld unless it can be shown that the board either abused its discretion or exceeded its statutory authority." NLRB v. Kaiser Agric. Chem., 473 F.2d 374, 382 (5th Cir. 1973). "The close relationship between labor policy and choice of remedy, coupled with the board's competence and expertise in the field of labor relations, dictate that the board's judgment [in fashioning a remedy] be given 'special respect by reviewing courts.'" Id. (quoting Gissel, 395 U.S. at 612 n.32). "And, 'it is for the Board not the courts to determine how the effect of prior unfair labor practices may be expunged.'" Franks Bros. Co. v. NLRB, 321 U.S. 702, 704 (1944) (quoting Int'l Ass'n Machinists v. NLRB, 311 U.S. 72, 82 (1940)).
Beyond a generic argument doubting the source of the NLRB's remedial authority, neither Tesla nor the plurality bother to articulate why they think Congress excluded from the NLRB's authority the ability to order deletion of a continuing, threatening, coercive, and therefore NLRA-violating tweet posted by a company's CEO. "Quite early on, the Court established that 'the relief which the statute empowers the Board to grant is to be adapted to the situation which calls for redress.'" Sure-Tan, Inc. v. NLRB, 467 U.S. 883, 900 (1984) (first quoting NLRB v. MacKay Radio &Tel. Co., 304 U.S. 333, 348 (1938); and then citing D. MCDOWELL &K. HUHN, NLRB REMEDIES FOR UNFAIR LABOR PRACTICES 8-15 (1976)). "The congressional determination to draft section 10(c) in indefinite language rather than to formulate preordained penalties for each offense [was designed to] allow[] the Board to set the tenor of its own authority by imaginative and specific treatment of the unique circumstances surrounding each unfair practice." See Dennis M. Flannery, The Need for Creative Orders Under Section 10(c) of the National Labor Relations Act, 112 U. PENN. L. REV. 69, 70 (1963). Given that the use of Twitter to commit labor violations is itself relatively unprecedented, the imaginativeness of the order to delete a tweet does not mean that the NLRB lacks the authority to order such a remedy.
Tesla spills significant ink questioning the source of the NLRB's remedial authority. The answer is a simple one. Congress gave the NLRB the authority (and the task) of "devising remedies to effectuate the policies of the Act" in 29 U.S.C. § 160(c). NLRB v. Seven-Up Bottling Co., 344 U.S. 344, 346 (1953). While Tesla and the plurality clearly disagree with Congress's policy-making decision to give the NLRB wide-ranging remedial authority, those kinds of policy judgments, under our separation of powers, come from Congress and the President, not judges.
Any suggestion that the NLRB's broad authority to fashion remedies excludes ordering Tesla to direct Musk to delete a threatening tweet (a permanently standing labor violation) would render the NLRB impotent to effectuate the NLRA in the face of employers' use of social media. Because there is no way to know precisely who saw or will see the permanent record of Musk's tweet, it would be left in place as a constant reminder to employees of the vulnerability of their stock options and perhaps other fringe benefits to unilateral recission by Tesla. The NLRB's deletion remedy remains sound even though Musk attempted to clarify days later that his earlier tweet was not intended by him or Tesla as a threat to take away employees' stock options. An NLRB remedy "must dispel, compensate for, or at least neutralize, the frustrating effects of persistent illegal activity." J.P. Stevens &Co. v. NLRB, 417 F.2d 533, 541 (5th Cir. 1969). The NLRB is not required to assume that after-the-fact assertions, buried down-thread or sprinkled in different Twitter threads, would have reached and changed the impression of every Tesla employee who read the earlier offending tweet prior to the election. Accordingly, the NLRB's remedy order is neither in excess of its broad statutory authority nor an abuse of discretion warranting remand under our highly deferential standard of review.
Second, there is no First Amendment issue posed by the deletion remedy because, contrary to the plurality's naked assertion, Musk's coercive tweet was not "constitutionally protected speech." Ante, at 8. While the First Amendment broadly states that "Congress shall make no law . . . abridging the freedom of speech," U.S. CONST. amend. I, the Supreme Court has never accepted the view that freedom of speech and association are "'absolutes.'" Konigsberg v. State Bar of Cal., 366 U.S. 36, 49 (1991). "[T]he First Amendment does not protect all speech, nor has it ever. 'There are certain well-defined and narrowly limited classes of speech, the prevention and punishment of which have never been thought to raise any Constitutional problem.'" Bailey v. Iles, 87 F.4th 275, 283 (5th Cir. 2023) (quoting Chaplinsky v. New Hampshire, 315 U.S. 568, 571-72 (1942)); cf. Herceg v. Hustler Mag., Inc., 814 F.2d 1017, 1020 (5th Cir. 1987) ("The Supreme Court has recognized that some types of speech are excluded from, or entitled only to narrowed constitutional protection.").
The plurality misrepresents FDRLST Media, LLC, 35 F.4th at 126, as vacating the NLRB's tweet deletion remedy because the remedy was violative of the First Amendment. Ante, at 8. The Third Circuit vacated the NLRB's tweet deletion order not because the remedy was improper, but rather because the "facetious and sarcastic tweet" at issue was not a labor violation in the first place. FDRLST Media, LLC, 35 F.4th at 127.
Relevant here, the Supreme Court has consistently held that the First Amendment does not protect threatening, coercive employer speech to employees in the labor organization election context-the precise category of speech Musk disseminated via Twitter. In NLRB v. Virginia Electric & Power Co., the Supreme Court took up a challenge to the NLRB's determination that the posting of an anti-union bulletin was a ULP from which the employer could be ordered to "cease and desist." 314 U.S. 469, 475-77 (1941). The Virginia Electric Court noted that the NLRA did not prohibit an employer "from expressing its view on labor policies or problems" unless the employer's speech "in connection with other circumstances [amounts] to coercion within the meaning of the Act." Id. at 477. By remanding for additional factfinding, the Court made clear that the mere presence of expression did not protect ULPs under the First Amendment: "The mere fact that language merges into a course of conduct does not put that whole course without the range of otherwise applicable administrative power. In determining whether the Company actually interfered with, restrained, and coerced its employees the Board has a right to look at what the Company has said as well as what it has done." Id. at 478. On remand, the NLRB (its membership by that point fully changed) found that the employer's "messages, whatever their significance when viewed standing alone, plainly cannot be regarded as mere expressions of the employer's opinion," and that "the posting of the bulletin was an integral part of the respondent's conduct, and as such, interfered with, restrained, and coerced the respondent's employees in the exercise of the rights guaranteed in Sec. 7 of the Act," a conclusion left undisturbed by subsequent appeals. Va. Elec. &Power Co., 44 N.L.R.B. 404, 428, 442 n.22 (1942), enforced by Va. Elec. &Power Co. v. NLRB, 132 F.2d 390, 396 (4th Cir. 1942), aff'd, 319 U.S. 533 (1943). Just four years later, in Thomas v. Collins, the Court framed Virginia Electric as "recogniz[ing] that employers' attempts to persuade to action with respect to joining or not joining unions are within the First Amendment's guaranty.... When to this persuasion other things are added which bring about coercion, or give it that character, the limit of the right has been passed." 323 U.S. 516, 537-38 (1945).
The principles announced in Virginia Electric and Collins were applied by the Court in NLRB v. Gissel Packing Company, 395 U.S. 575 (1969). At issue in Gissel were an employer's written and spoken anti-union statements. 395 U.S. at 619. In considering a First Amendment challenge, the Supreme Court emphasized that an employer's "free speech right to communicate his views to his employees" is "firmly established." Id. at 617. Still, any assessment of employer expression "must be made in the context of its labor relations setting," where "an employer's rights cannot outweigh the equal rights of the employees to associate freely." Id. "[A]ny balancing of those rights," the Court explained, "must take into account the economic dependence of the employees on their employers, and the necessary tendency of the former, because of that relationship, to pick up intended implications of the latter that might be more readily dismissed by a more disinterested ear." Id. The Court went on to hold that an employer's coercive speech falls "outside the protection of the First Amendment and 8(c)." Id. at 579.
Further still, in Chamber of Commerce v. Brown, 554 U.S. 60 (2008), relying on Virginia Electric, Collins, and Gissel, the Court again found that free debate is not without limits. In that case, the Court found that California laws regulating employer speech within a zone protected and reserved for market freedom were preempted by the NLRA. Relevant to Musk's tweet, in laying out its preemption analysis, the Court emphasized that the NLRA "implements the First Amendment" and demonstrates "congressional intent to encourage free debate on issues dividing labor and management." Id. at 66-67 (citations omitted). The Supreme Court found that only noncoercive employer speech about unionization is protected by the First Amendment. Id. And our own sturdy precedent conforms with these holdings. See Riley-Beaird, Inc., 681 F.2d at 1086 (recognizing that the issue of whether speech is protected under the First Amendment is the same as whether it constituted a threat in violation of the NLRA-because the First Amendment does not protect that type of speech-so, "the issue before us is not where to draw the line separating protected from unprotected speech. That line is settled. The only issue is whether the Board correctly determined that [Musk]'s conduct constituted a violation of section 8(a)(1)").
Together these cases show that the First Amendment protects to a great degree an employer's right to speak out against union representation of its employees, but the First Amendment does not grant employers carte blanche to use their speech to commit labor violations. For the reasons stated earlier, substantial evidence supported the NLRB's finding that Musk used his Twitter account to threaten to unilaterally deprive unionizing employees of stock option benefits. Even the plurality "assume[s] without deciding that Musk's tweet violated the NLRA." Ante, at 7. Faithfully applying Virginia Electric, Thomas, Gissel, and Chamber of Commerce, which we must as an inferior appellate court, the First Amendment does not protect Musk's coercive speech, regardless of whether it is uttered on Twitter, in a newspaper, or on the factory floor. See, e.g., NLRB v. Laredo Coca Cola Bottling Co., 613 F.2d 1338, 1341 (5th Cir. 1980) (affirming the NLRB's finding of retaliatory threat that employer delivered in the form of quotation to a newspaper); Citizens United v. Fed. Election Comm'n, 558 U.S. 310, 326 (2010) (holding that "[w]e must decline to draw, and then redraw, constitutional lines based on the particular media or technology used" by a particular speaker). Given that Musk's coercive speech was unprotected, the NLRB's subsequent tweet deletion order-a remedy that has ample historical precedent-poses no First Amendment issue. See LAURENCE H. TRIBE, AMERICAN CONSTITUTIONAL LAW (2d ed. 1988) § 12-37, pp. 1054-1055 ("Once specific expressional acts are properly determined to be unprotected by the first amendment, there can be no objection to their subsequent suppression or prosecution.").
See, e.g., Alexander v. United States, 509 U.S. 544, 550 (1993) (authorizing a court order to seize and destroy books, magazines, and films from a person convicted of an obscenity law without any finding that those materials were obscene or otherwise unprotected by the First Amendment); Balboa Island Vill. Inn, Inc. v. Lemen, 40 Cal.4th 1141, 1155-56, 156 P.3d 339, 349 (Cal. 2007), as modified (Apr. 26, 2007) (holding that a trial court had authority to require an individual Yelp user to delete speech on the platform because the speech was defamatory and, thus, unprotected by the First Amendment); Kinney v. Barnes, 443 S.W.3d 87, 93 (Tex. 2014) (permitting a court order "to remove [defamatory] statements from . . . websites . . . upon a final adjudication that the statements are defamatory"); Glennon v. Rosenblum, 325 F.Supp.2d 1255, 1269 (N.D. Ala. 2018) (ordering websites carrying a defendant's defamatory story to delete it from their internet platforms). The plurality says that only "non-speech material" can be destroyed. Ante, at 7. But the case cited for this proposition rejects precisely this "simplistic, all-or-nothing-at-all approach to First Amendment protection." R.A.V. v. City of St. Paul, 505 U.S. 377, 384 (1992). R.A.V. explained that certain "areas of speech can, consistently with the First Amendment, be regulated because of their constitutionally proscribable content (obscenity, defamation, etc.)-not that they are categories of speech entirely invisible to the Constitution." Id. at 383 (emphasis in original). As with fighting words or true threats, remediation of employer coercion under the NLRA is permissible because-though it may travel through the channels of speech-such coercion is unprotected as an "element of communication." See id. at 386. It strains credulity to suggest that the Virginia Electric Court would have viewed the NLRB's cease and desist order to be unconstitutional if it specifically required the company to take down intimidating bulletins in the workplace rather than refrain from posting them. Does an employer or a proprietor or a landlord have similar First Amendment immunity when posting a notice that those of the wrong sex or race or age have no place on its premises, so long as it does so before any of the civil rights laws are enforced against it? The plurality's unprecedented First Amendment simplification aside, it is not clear to me that this case would even fall within the ambit of such a dubious rule. The recirculation of messages on Twitter is "the product of a wealth of choices about whether- and, if so, how-to convey posts," distinct from the speaker's own expression in the first instance. Moody v. NetChoice, LLC, 144 S.Ct. 2383, 2405 (2024). Such continuing recirculation on Twitter's public interface, not any specific act of expression stored on Twitter's servers, is what is targeted here. See, e.g., Pittsburgh Press Co. v. Pittsburgh Comm'n on Hum. Rels., 413 U.S. 376, 388 (1973) ("Discrimination in employment is not only commercial activity, it is illegal commercial activity under the Ordinance. We have no doubt that a newspaper constitutionally could be forbidden to publish a want ad proposing a sale of narcotics or soliciting prostitutes.... The illegality in this case may be less overt, but we see no difference in principle here." (footnote omitted)). Options may be available to the NLRB that address the first imperative without running afoul of the plurality's view of the second. Because we remand, the NLRB may consider potential remedies that would allow posts to be preserved (which is to say, not "deleted") without continuing their injurious recirculation, such as allowing a post to be downloaded to a private archive or copied to a private account before removal, or enabling some posts to be made private while an account is otherwise maintained as public.
The plurality invokes the adage of counter-speech, arguing that "'the remedy' for bad speech, after all, should be 'more speech, not enforced silence.'" Ante, at 6-7 (quoting Whitney v. California, 274 U.S. 357, 377 (1927) (BRANDEIS, J., concurring)). The plurality goes on to concede that deletion can be a proper remedy for speech that falls outside of the First Amendment's protection. Id. at 7. It is precisely because Musk's coercive speech is unprotected that the NLRB's deletion order does not offend the First Amendment.
V
Finally, I would uphold the NLRB's finding that Tesla terminated Ortiz for not divulging information about protected union activities during an interrogation in violation of Section 8(a)(1) and (a)(3) of the NLRA.
Section 8(a)(3) of the NLRA makes it a ULP for an employer to discriminate "in regard to hire or tenure of employment . . . to encourage or discourage membership in" a union. See 29 U.S.C. § 158(a)(3). Such discrimination interferes with Section 7 rights, so it also violates Section 8(a)(1). See Metro. Edison Co. v. NLRB, 460 U.S. 693, 698 n.4 (1984); 29 U.S.C. § 158(a)(1). Thus, "an employer violates section 8(a)(3) and (1) of the Act by discharging employees because of their union activity." NLRB v. ADCO Elec. Inc., 6 F.3d 1110, 1116 (5th Cir. 1993). While an employer has a legitimate right to investigate "facially valid complaints of employee misconduct, including complaints of harassment," the NLRA also protects an employee's right to keep his or her union activities confidential, even if that means giving evasive or untruthful answers in response to an employer's questions that an employee reasonably believes are inquiries into protected union activity. Fresenius USA Mfg., Inc., 362 N.L.R.B. 1065, 1065 (2015); Paragon Sys., Inc. &Arthur J. Blake, 362 N.L.R.B. 1561, 1565 (2015); Consol. Diesel Co. v. NLRB, 263 F.3d 345, 353-54 (4th Cir. 2001).
The NLRB balances those two rights through the following rule: If an employee has a "reasonable basis" for believing that her employer is attempting to uncover protected union activity, and if the questioning employer has no legitimate business justification for doing so, then an untruthful or evasive employee response concerning that activity is not a lawful basis for discharge. See, e.g., Tradewaste Incineration, 336 N.L.R.B. 902, 907 (2001); St. Louis Car Co., 108 N.L.R.B. 1523, 1525-26 (1954); Paragon Sys., Inc., 362 NLRB at 1565; accord, e.g., Onyx Env't Servs., L.L.C., 336 N.L.R.B. 902, 907 (2001). One way our court has measured the "reasonableness" of an employee's belief is by asking whether the employee's false or "evasive statement[] . . . in response to [employer] questioning [is] 'inextricably involved' with the employee's protected conduct." Cordua, 985 F.3d at 429 (quoting NLRB v. Roney Plaza Apartments, 597 F.2d 1046, 1051 (5th Cir. 1979)). The NLRB's balancing rule recognizes employees' "normal reluctance to divulge" protected activity to employers, who often use such information to retaliate. Aladdin Indus., Inc., 147 N.L.R.B. 1392, 1407 (1964). The rule also aligns with the well-established principle that untruthful answers to questioning about union activity are evidence of employer coercion-not employee misconduct. See Sturgis Newport Bus. Forms, Inc. v. NLRB, 563 F.2d 1252, 1256 (5th Cir. 1977); Jays Foods, Inc. v. NLRB, 573 F.2d 438, 444 (7th Cir. 1978). If such answers justified discharge, an employer could easily rid itself of union supporters by asking questions about protected activity likely to provoke evasion or untruthfulness.
Applying these standards here, Ortiz's false statement-responding "I don't know" to Gecewich asking where the Workday screenshots came from-is "inextricably involved" with Ortiz's protected conduct. The NLRB, in adopting the ALJ's finding, described Ortiz's protected concerted activity as all of the actions that Ortiz and Moran took "[u]pon learning that employees testified on behalf of Tesla during a union-sponsored California State Assembly bill[.]" This includes: (1) Ortiz asking Moran to help him learn if these individuals were current employees; (2) Moran then searching Workday system to search for these employees; (3) Moran sending the screenshots of their Workday profiles to Ortiz; and (4) Ortiz then posting the screenshots with his comments on the union employees' private Facebook page. Tesla concedes that it interviewed Ortiz about these protected union activities, during which interview Ortiz lied about knowing the source of the screenshots, the stated reason for Ortiz's termination. Ortiz's false statement in response to Gecewich's question obviously meets the "inextricably involved" standard set by Roney Plaza Apartments, 597 F.2d at 1051, and reaffirmed by Cordua, 985 F.3d at 429. It follows, then, that Ortiz "reasonably understood that Gecewich was trying to learn about his protected activities when [Gecewich] repeatedly asked who sent him the Workday profile screenshots." Ortiz was "scared to death" that he would be discharged and did not want to get Moran in trouble. And for good reason: that is exactly what happened-Ortiz was fired, and Moran was disciplined.
Tesla does not dispute this NLRB finding and, therefore, it has waived the issue. See Flex Frac Logistics, L.L.C., 746 F.3d at 208.
Nor did Tesla have an alternative lawful motive to question Ortiz. Tesla argues that it was faced with a facially valid complaint of misconduct. "[O]ne of the employees targeted in the Facebook post (Pratt) submitted it to a member of the HR department (Hedges) and said he felt harassed and targeted by Ortiz." That report eventually made its way to Gecewich, who apparently "understood that Pratt felt targeted and harassed" because of the Facebook post. Tesla relies on the precept that an employer may question an employee about their protected union activity when the employer has a "legitimate business justification" for doing so, United Servs. Auto. Ass'n v. NLRB, 387 F.3d 908, 916 (D.C. Cir. 2004), and the precept that "employers have a legitimate business interest in investigating facially valid complaints of employee misconduct, including complaints of harassment." Fresenius USA Mfg., Inc., 362 NLRB at 1065.
Substantial evidence supported the NLRB's factual finding that Tesla's questioning of Ortiz was not part of a facially valid complaint of employee misconduct. Specifically, the ALJ found that the initial complaint of harassment from Pratt to Hedges, the Tesla human resources official who had invited Pratt to testify against the UAW legislation, and Hedges's complaint to Gecewich, was disingenuous given the content of Pratt's text messages to Hedges. "Pratt complained that Ortiz' Facebook post made him feel singled out, but this claim [wa]s disingenuous since he forwarded the Facebook post to Hedges with the remark, 'Looks like we got under someone's skin' with a smiling face and eyes and rosy cheeks emoji. This addition of the emoji does not reflect a concern of harassment." When Hedges responded by asking if the post was on Facebook, Pratt replied "Yea lol." Gecewich was shown a copy of this text exchange. Pratt later explained to the investigator, according to the investigator notes, that he sent the post to Hedges "more as a we are getting a rise out of people."
The ALJ further based its finding that Pratt's claim of harassment was not a facially valid complaint of employee misconduct on: (1) other ULPs established in the record (including a prior instance of Tesla improperly interrogating Ortiz about his union activities); (2) Gecewich's knowledge of the context within which the Facebook post and Workday use occurred; and (3) Gecewich's questioning of Ortiz, which was aimed at discovering the identity of Ortiz's source and fellow pro-union colleague, Moran- information that Tesla did not otherwise have a legitimate right to discover, considering that Tesla had no rule restricting employee use of Workday. In fact, it is undisputed that Tesla had no rule against employees accessing Workday at the time it began its investigation into Ortiz, and Tesla does not contest the NLRB's determination that it violated the NLRA by disciplining Moran for accessing Workday in this manner. All of this is substantial evidence that Tesla had no legitimate business justification to question Ortiz about his protected union activity. "A court reviewing an agency's adjudicative action should accept the agency's factual findings if those findings are supported by substantial evidence . . . [and] should not supplant the agency's findings merely by identifying alternative findings that could be supported by substantial evidence." Arkansas v. Oklahoma, 503 U.S. 91, 11213 (1992).
Considering that Ortiz "reasonably understood that Gecewich was trying to learn about his protected activities" and the substantial evidence that Tesla had no legitimate business justification to question Ortiz about the same, the NLRB reasonably found that Ortiz had no obligation to respond truthfully. Ortiz's "dishonesty"-that he could not recall who sent him the Workday screenshots-does "not constitute a lawful reason to discharge h[im]." United Servs. Auto. Ass'n, 387 F.3d at 916-17 ("There is, then, substantial evidence in the record to support the Board's finding that the employees could reasonably believe that the company had only one objective in questioning Williams and the other employee: to identify, with certainty, who had engaged in the protected concerted activity." (citations omitted)). Accordingly, Tesla's assertion that it discharged Ortiz for concealing his coworker's union activity constitutes an admission of unlawful motive under settled law. See ADCO Elec. Inc., 6 F.3d at 1116 ("[A]n employer violates section 8(a)(3) and (1) of the Act by discharging employees because of their union activity.").
The en banc plurality has no rebuttal to this analysis. Instead, in another punt, it vacates the NLRB's Ortiz reinstatement order and remands for the NLRB to consider that the ultimate decisionmaker, Graminger, was pro-union, falsely claiming that the NLRB failed to do so the first time around. Ante, at 9-11. That is error for three independent reasons.
First, that evidence is immaterial to the conclusion that Ortiz was not required to answer Gecewich's question truthfully, and, again, Tesla and Graminger admit they terminated Ortiz because of his "dishonesty" to Gecewich. That Graminger was purportedly pro-union did not magically give him a lawful reason to terminate Ortiz when Ortiz was fired for declining to divulge information about protected union activities during an interrogation.
Second, the ALJ and NLRB took Graminger's testimony "into account." Universal Camera Corp. v. NLRB, 340 U.S. 474, 488 (1951). The ALJ expressly found that Graminger was "more credible" than the other witnesses for Tesla. This shows the ALJ considered his testimony, even though it did not specifically mention his pro-union beliefs. The ALJ expressly did not credit Graminger's testimony that Ortiz's union activity was never discussed during the meeting about Ortiz because it was "obvious that all the attendees of the meeting knew that Ortiz was active with the Union, and it seem[ed] implausible that no one mentioned his union activity during this meeting especially considering Graminger admitted this was a 'sensitive case.'" Graminger testified that he asked a senior vice president (Hochholdinger) whether he knew of the internal investigation "involving Richard Ortiz, a member of the union." While Hochholdinger confirmed that there had been a similar instance where someone was terminated for lying during an investigation, the ALJ noted that Graminger "did not know which Tesla policies Ortiz violated nor did he review any policies" when he recommended Ortiz's termination. The record shows that the ALJ fully considered Graminger's testimony, identified inconsistencies within his testimony (specifically concerning discussions of Ortiz's union activities), and concluded Ortiz's termination was motived by anti-union animus. The NLRB reviewed the entire record and agreed. Remanding for the NLRB to reconsider Graminger's testimony second-guesses the NLRB's credibility determination and reweighs the evidence, which we are not permitted to do. Allied Aviation Fueling, 490 F.3d at 378 ("In determining whether the Board's factual findings are supported by the record, we do not make credibility determinations or reweigh the evidence.").
Tesla attempts to paint its decision to fire Ortiz in a positive light; specifically, it characterizes its decision as an effort to treat Ortiz the same as any other Tesla employee. The devil is in the details. Tesla only previously fired an employee for dishonesty where that employee, a vice president, lied about his use of a company vehicle, drugs, and alcohol found in that vehicle (i.e., a misuse of company resources) as well as an improper relationship with another employee. Those facts are worlds apart from Ortiz responding "I don't know" when asked who provided him with screenshots containing information from Workday that Tesla did not prohibit employees from accessing, especially when Ortiz held a reasonable belief that Tesla was unlawfully interrogating him to learn about his and others' protected union activities.
Third, Graminger made his decision to terminate Ortiz in reliance on incomplete and inaccurate facts reported to him by Gecewich, who the ALJ found to be wholly incredible. For example, Gecewich failed to inform Graminger of the very impetus of Ortiz's and Moran's conduct: Pratt's and Ives's public testimony on behalf of Tesla to state legislators in opposition to a union-sponsored bill. And Gecewich lied to Graminger that Ortiz's Facebook post had "leaked some internal information out of Workday including some telephone number and personal information." Because these material omissions and misrepresentations tainted Graminger's ultimate decision, Tesla may not wash away the substantial evidence of anti-union animus by pointing to Graminger's neutrality. NLRB v. Neuhoff Bros., Packers, Inc., 375 F.2d 372, 374-75 (5th Cir. 1967); accord NLRB v. Big Three Indus. Gas &Equip. Co., 579 F.2d 304, 312 (5th Cir. 1978); Charter Commc'ns, Inc. v. NLRB, 939 F.3d 798, 816 (6th Cir. 2019).
* * *
Because our en banc court should have enforced the NLRB's orders across the board, I would deny Tesla's petition for review and grant the NLRB's cross-application to enforce its order. I respectfully dissent from the plurality's opinion and judgment.
[*]JUDGE HO is recused and did not participate in this decision.