Every day millions of Americans publish opinions on the Internet; if they do so unnamed or by pseudonym, they can expect to communicate anonymously. Every day prosecutors issue subpoenas to determine whether crimes have been committed; if they do so in reasonable good faith, they can expect those subpoenas will be enforced. These expectations collided last year in In re Grand Jury Subpoena (Glassdoor), No. 17-16221 (9th Cir. Nov. 8, 2017), a closely watched appeal in the Ninth Circuit from an order compelling the operators of an employee-review website to disclose identifying information of users posting anonymous comments about a company subject to grand jury investigation.

The panel upheld the district court, rejecting First Amendment challenges interposed (on users’ behalf) in a motion to quash. The opinion has drawn thoughtful commentary, regarding the tension between the public’s interest in the prompt investigation of crime and the right to anonymous expression. Although litigation over anonymous online speakers is unlikely to be settled by this controversy alone, there are reasons to expect (or hope) that prosecutors will pursue other avenues to locate persons with knowledge of corporate fraud, and that the compelled outing of commentators might remain an unusual last resort.


The controversy arose from an investigation by the Arizona U.S. Attorney into the operations of a government contractor administering two VA healthcare programs. The government served a subpoena on Glassdoor, operating the “Glassdoor.com” website, where employers promote job openings and employees rate the companies on working conditions and benefits. Reviewers post to the site anonymously. The subpoena sought every “company review” regarding the subject contractor, together with names, addresses and identifying data of the reviewers. There were 125 postings, but the government agreed after discussions to limit its request to eight reviews, critical of the contractor, that were attached to the subpoena.

Despite the government’s trimming, Glassdoor moved to quash based on its users’ rights to anonymous speech. The district court denied the motion, relying on Branzburg v. Hayes, 408 U.S. 665 (1972), in which the U.S. Supreme Court held that reporters must cooperate with a grand jury, in spite of anonymity promises to sources, where the government probe is not undertaken in bad faith. Glassdoor had argued that the Branzburg “bad faith” test was inapposite, insisting that the Ninth Circuit’s holding in Bursey v. United States, 466 F.2d 1059 (9th Cir. 1972), set a stricter standard of review.

Which precedent should control was again the battleground on appeal. Glassdoor argued that the rights of anonymous speakers, see McIntyre v. Ohio Elections Commission, 514 U.S. 335 (1995), were distinct from the reporter’s privilege claims in Branzburg. They urged instead that Bursey’s three-part test controlled, requiring an immediate and substantial interest of the government in the information; a substantial connection between the information and that interest; and the least drastic means necessary to secure it. Bursey, 466 F.2d at 1083.

The panel rejected Glassdoor’s Branzburg analysis as a “distinction without a difference,” noting that the Supreme Court had there addressed the argument that an inability on the part of journalists to safeguard confidential sources would deter reluctant speakers, compromising the free flow of information. Thus Glassdoor, a forum for anonymous speech and would-be guarantor of that anonymity, echoed the Branzburg journalists. The panel also emphasized that the site’s anonymous-use policy had caveats. Users were required to certify their acceptance of Glassdoor’s terms, including a “Privacy Policy” qualifying its general non-disclosure to permit compliance with “relevant laws … subpoenas … or legal process[.]” (The Circuit also disagreed that enforcing the subpoena would infringe users’ “associational privacy,” reasoning that the website, comprising anonymous commentary among participants unknown to one another, did not implicate the “expressive association” protected by NAACP v. Alabama, 357 U.S. 449 (1958).)


The scope of this article cannot fully digest the serious issues implicated, and readers should review the panel opinion and briefing, including amici curiae submissions by Electronic Frontier Foundation and others. A few observations about the panel’s approach:

Terms of use disclaimers inconsistent with absolute anonymity. Speech advocates fairly complain that the practicalities of online forums require some identification for the user to participate. The panel stressed this pedigree data, and the site’s disclaimer about response to legal process, to question whether users had an absolute expectation of privacy. Although such “fine print” could be seen as expanding compelled disclosure, Glassdoor did fight the good fight for its users here, and intrusions on anonymity should still be exceedingly rare. Civil subpoenas, for instance, would not benefit from the Branzburg analysis employed. (Moreover, though it may be small comfort to advocates, Rule 6(e) grand jury secrecy might still conceal from the public the identity of speakers approached by investigators.)

‘Branzburg’ and ‘Bursey’ analysis perhaps not that different. Although the panel came down in favor of Branzburg and the government, it stated that it would have reached the same outcome under Bursey. The latter case involved a probe of the Black Panthers newspaper, related to alleged threats against President Nixon. The investigators’ transgression there was in the second Bursey prong—the lack of a “substantial connection” between the legitimate interest in investigating threats and the political subjects actually scrutinized. That disconnect was the constitutional grievance, and it is hard to distinguish that intrusion from the “bad faith” exception recognized in Branzburg. Under either approach, deference to grand juries would seem to extend only to genuine pursuit of criminal evidence.

No ultimate privilege against users’ testimony. The panel noted in its Branzburg analysis that the court declined to create a “privilege” for reporters to avoid a grand jury where the protected sources themselves would enjoy no such privilege—the Fifth Amendment being the only exception to the rule that the public has a right to every citizen’s evidence. This would seem to be the sticking point in any confidentiality confrontation. If a potential witness possesses evidence owed the public, and there is a means to compel appearance, no intervening claim of privilege (other than self-incrimination) will likely prevail.

To Cast a Wide Net?

That the Glassdoor panel endorsed the government’s request for users’ identities—at least as to the shorter list of eight—is not surprising. Courts do defer to grand jury investigations, and absent truly intrusive overreaching motions to quash will seldom prevail. What would be at risk in civil litigation of rejection by a court as a “fishing expedition” will often merit enforcement through the grand jury, because of the clear societal interest in investigating potential crimes. The interesting question posed by scenarios like Glassdoor, then, may not be whether the government can go fishing, but whether it should. With apologies to Matthew’s Gospel, a broad subpoena for unknown persons will make prosecutors “fishers of men,” and there are reasons to consider what kind of catch you might pull before you put your net in the water.

Anonymity can be a powerful catalyst for truth, but it can also be a haven for exaggeration, vindictiveness and outright falsehood. In the employment context, the urge to “flame” the company can be considerable and, on the nameless Internet, usually without consequence. Anonymous commentators who exaggerate or invent wrongdoing can create headaches for the government when approached by agents. If one lies, and defends his inaccurate posting, he has not only misled investigators but likely committed a crime that could explode in his face (and the government’s case). If one tells the truth, and backs away from overheated accusations to deny fraud, her account could well be exculpatory material subject to disclosure under the Brady doctrine. Some posts would be accurate and corroborated, but the bigger the net the greater the risk, and the potential problems are both serious and avoidable.

As tempting as it might be to pursue sources of unsigned posts, the government already has the ability to survey employees from whom such comments would have come. With organizational charts and employee lists, and assisted by company counsel, prosecutors can identify persons with knowledge. Represented by counsel in interviews, employees should be well prepared to give coherent, truthful accounts of whatever they know. A meeting in a prosecutor’s office is an experience most will take seriously; a visit to the internet may not be.

It is also worth remembering that a worthy insider witness—an organized thinker who wants to do the right thing—should have the wherewithal to contact the government if so motivated. Statutory incentives like the False Claims Act will attract some, but the larger question is whether it makes sense to pursue a source who sounds an “alarm” without stepping forward. (In negotiations preceding its motion Glassdoor apparently made some offer to contact users to explore voluntary cooperation, but the prosecutors declined. That is perhaps understandable, but such compromise merits consideration. Except where the information implicates physical safety, it is difficult to see any real return in chasing non-volunteers, and reticence may have its reasons.)


Online speech is abundant and often unattributed. The Glassdoor ruling suggests that where content hints at knowledge of criminality investigators will be permitted to track its sources. The temptation to peel back the cover is obvious, but heedfulness is warranted—to avoid chilling salutary speech, and to escape grief from the more mischievous kind.

Michael Grudberg is a partner at Tarter Krinsky & Drogin.