The Volokh Conspiracy

Mostly law professors | Sometimes contrarian | Often libertarian | Always independent

Free Speech

Someone Trying to Vanish My Article About People Trying to Vanish Articles


In 2021, I published Shenanigans (Internet Takedown Edition) in the Utah Law Review; the article is about people using various schemes—some of which included forgeries and frauds—to get material vanished from Internet search results. (You can read the Introduction below.)

Yesterday, I saw that someone tried to use a different scheme, which I briefly mentioned in the article (pp. 300-01), to try to deindex the Utah Law Review version of my article: They sent a Digital Millennium Copyright Act notice to Google claiming that they owned the copyright in my article, and that the Utah Law Review version was an unauthorized copy of the version that I had posted on my own site:

(That's a copy of the takedown demand that's archived at the Lumen Database.) Google has apparently recognized that this demand is unfounded, and hasn't blocked access to the "allegedly infringing URL[]." Too bad, Liam.

Here's the Introduction to my article, in case you're interested; you can also read the PDF here. Note that I have no reason to think that the DMCA notice stems from the particular case with which the Introduction begins; that's just one of the many fraudulent takedown requests that I discuss in my article.

[* * *]

In 2016, Google received a copy of a Miami-Dade County default judgment in MergeworthRX, Inc. v. Ampel, No. 13-13548 CA. A certain web page, the judgment said, was libelous:

2. The reports posted on or about December 30, 2014 by Defendant, CELIA AMPEL on regarding Plaintiffs, MERGEWORTHRX, INC. and STEPHEN CICHY (the "Report"), which is available at contains defamatory statements regarding Plaintiffs.

The submitter, therefore, asked Google to "deindex" that page—remove it from Google's indexes, so that people searching for "mergeworthrx" or "stephen cichy" or "anthony minnuto" (another name mentioned on the page) wouldn't see it.

Google often acts on such requests, as it did on this one, effectively vanishing the material from the Internet. And why not? It's a service to Google's users, who presumably want to see true information, not information that's been found libelous. It's good for the people who were libeled. It can let people at Google feel that they are doing good. And it's respectful of the court judgment, even though it's not strictly required by the judgment. Win-win-win-win.

Except there was no court order. Case No. 13-13548 CA was a completely different case. Celia Ampel, a reporter for the South Florida Daily Business Review, was never sued by MergeworthRX. The file submitted to Google was a forgery.

It was one of over 90 documents submitted to Google (and to other hosting platforms) that I believe to be forgeries. Google's well-meaning deindexing policy has prompted a rash of such apparent forgeries, some seemingly home brewed and some done for money as part of a "reputation management company" business model. Such reputation management, whether fraudulent or legitimate, is big business.

And those over 90 items are just the apparent forgeries, which are possible for Google to spot. Google seems to check most of the submissions it gets against court records, many of which are available online (as the Miami-Dade County records are). Most such apparent forgeries, I think, are identified as suspect and thus ignored by Google—though, a few, such as the MergeworthRX forgery, do get acted on.

But what if a reputation management company engineers a real lawsuit involving a fake defendant? It sends the court a complaint, purportedly from the plaintiff, and an answer admitting liability and stipulating to a judgment, purportedly from the defendant. The court is generally happy to accept the apparent stipulation, enter the injunction, and get the case off its docket—having no idea, of course, that the defendant doesn't really exist. And Google can't easily recognize that the defendant doesn't really exist, either. I have found about 30 cases that seem to fit this pattern.

Or what if such a company engineers a real libel lawsuit involving a real defendant—but one who has nothing to do with the allegedly libelous post? The company again sends the court a complaint and an answer with a stipulation, and the answer and stipulation are signed by the real defendant; indeed, the defendant's signature is even notarized.

It's just that the stipulation that the defendant authored the post and admitted that it's false is itself false. But again, the court doesn't know, so it issues the injunction, and Google doesn't know that, either. I have found what appear to be over 30 of those, though here the evidence is less open and shut.

Or what if the plaintiff doesn't try hard to find the defendant, but instead gets authorization to serve the defendant by publication (which the defendant is nearly certain never to learn about), and then gets a default judgment when the defendant doesn't show up? In normal lawsuits, where the point of the judgment is to get damages or force the defendant to do something, defendants could successfully move to set aside the defaults on the grounds that the defendants hadn't been properly served with the lawsuits. But the point of these particular lawsuits is not to get the defendants to do something: It's to persuade Google to do something, and Google has no idea whether the plaintiffs had done a good enough job of finding the defendants. I have found over 60 cases like this, though here, too, the evidence is less clear.

And there's more: some orders, for instance, were obtained against people who wrote comments attached to mainstream media articles but were then submitted to Google in an attempt to deindex the whole article, though there is no evidence that the underlying article is libelous. Indeed, some of the comments might have actually been planted by a reputation management company precisely as an excuse to justify the lawsuit.

Some other orders include the URLs of government documents or of newspaper articles buried in a long list of URLs that were supposedly written by the defendant, even though there's no reason to think the defendant posted those documents. Still others use alleged confessions by defendants quoted in newspaper articles as a tool for trying to vanish the article as a whole.

In all, I found about 700 seemingly legitimate U.S. libel case orders submitted to Google for takedown and then forwarded to the Lumen Database from 2012 up to mid-October 2016. I also found, from the same date range:

  • Over 90 apparently forged orders (my total forgery count includes some post-October 2016 submissions),
  • over 30 possible fake-defendant cases,
  • likely over 30 apparently fake-claim-of-authorship cases,
  • about 10 cases aimed at deindexing government documents or newspaper articles, which were undoubtedly not written by the defendant, and
  • about 60 cases in which there seemed to be no real attempt to track down and serve the defendant.

That's a total of over 200 either obviously forged or fraudulent or at least highly suspicious cases. And it's hard to tell how many of the 700 seemingly legitimate orders might also have involved various kinds of frauds that were just too subtle to catch. I'm sure there are many perfectly proper libel judgments that lead to deindexing requests. But many deindexing orders are suspect in various ways. I go through the details of these shenanigans—and others—in the pages below, and then turn to a few implications.

[1.] Most obviously, this is a reminder not to trust apparent court orders until you check them against court records. Say, for instance, you run a website, and someone sends you an order declaring certain material on the site to be libelous—an order addressed not to you, but to one of your users—and asks you to take down the alleged libel. Don't assume the order is legitimate: Verify whether it actually appears in the court docket, and, if possible, check with the author of the material (if you can identify and reach the author) to see if the order may have been fraudulently obtained.

[2.] Likewise, if you are a lawyer and a prospective client asks you to file prepared documents—such as stipulations or affidavits—you should be careful. If the documents are fraudulent, you might end up tainted by the fraud, even if you hadn't known about it.

[3.] This is also a reminder of the value of American courts' policy of keeping records presumptively open, even when the records involve relatively small-time low-profile litigation. If such records were presumptively sealed, fraud would be even easier to hide.

[4.] Sometimes some orders are sealed—very rarely in libel cases, but often in criminal expungement cases. People whose records have been expunged sometimes send those expungement orders to Google and to others to try to get mugshots, newspaper police blotter entries, and even normal newspaper articles deindexed or edited. To my knowledge, Google doesn't act on such expungement orders, but some newspapers and other sites might be willing to.

Yet, if the order is sealed—or if the file is entirely destroyed, which is what happens in some jurisdictions—the recipient can't readily confirm that the order is authentic. The most it can get from the court system is "that file is sealed." If courts do want to operate using sealed expungement orders, then they need to create a verification system where someone who has a copy of what purports to be such an order can confirm whether that order is identical to the one in the court file.

[5.] Some anti-libel injunctions aim to bind search engines, hosting companies (such as WordPress or Yelp), domain name registrars, and the like—even when those entities were not parties to the case. In its fractured 3–1–3 decision in Hassell v. Bird, the California Supreme Court recently held that such injunctions are impermissible.

The magnitude of possible frauds in such cases helps show that the Hassell plurality and the concurrence were correct: Online service providers, such as Yelp and Google, that get these orders are the first line of defense against fraudulent behavior, so long as they have no legal obligation to comply with such orders issued against third parties. The service providers can exercise their discretion to conclude that some orders appear untrustworthy. They can demand more documentation from people who submit the orders. And if their concerns about the orders are not adequately resolved, they can decline to act on the orders.

[6.] Congress is considering whether to strip Internet platforms of libel imunity if they fail to take down material after being sent

[a] copy of the order of a Federal or State court under which the content or activity was determined to violate Federal law or State defamation law, and to the extent available, any references substantiating the validity of the order, such as the web addresses of public court docket information.

But the cases I discuss below show that such orders are not trustworthy: Even if they aren't forged, there is no way of figuring out if they are the result of some other fraud—or are otherwise unreliable, for instance because they have been obtained as a result of a default judgment, with no real attempt to locate and serve the defendant. The orders may purport to "determine[]" that certain "content" "violate[s] … defamation law," but that determination, even if fair as between the parties, ought not be used to determine the rights and responsibilities of third parties who were not involved in the litigation.