China's New Internet Regulations, Building On Western Internet Regulations, Requires Algorithms To 'Vigorously Disseminate Positive Energy'

from the justifying-actual-suppression-of-ideas dept

When the UK announced its rebranded "Online Safety" bill (originally, the "Online Harms" bill) we noted that the mechanism included was effectively identical to the original Great Firewall of China. That is, when China first began censoring its internet, rather than telling websites explicitly what needed to be taken down, it just gave vague policy guidance about what "harmful" information would be a problem if it was found online, and backed that up with a serious threat: if any service provider was found not to have taken down information the government deemed problematic, it would face serious consequences. There was, of course, no such threat for taking down information that should not have been taken down. The end result was clear: when in doubt, take it down.

It remains preposterous to me that all across Western democracies, we've seen them taking the same basic approach -- insisting that platforms need to be much more aggressive in pulling down "bad" information (loosely defined), with significant liability attached to leaving it up (even if the content is legal), and little in the way of punishment for overblocking. And, over and over and over again studies have shown that when you set up a liability regime this way, you get massive overblocking. It seems that some countries see that as a feature, not a bug.

And, of course, with that approach now being picked up in other countries, China has apparently decided to ramp up its own approach. Under the guise of stopping "harmful" information online, China has released a draft of new regulations to punish internet companies that don't remove harmful information:

China kicked off a two-month campaign to crack down on commercial platforms and social media accounts that post finance-related information that’s deemed harmful to its economy.

The initiative will focus on rectifying violations including those that “maliciously” bad-mouth China’s financial markets and falsely interpret domestic policies and economic data, the Cyberspace Administration of China said in a statement late Friday. Those who republish foreign media reports or commentaries that falsely interpret domestic financial topics “without taking a stance or making a judgment” will also be targeted, it added.

Part of this new effort is a set of draft regulations for "algorithmic recommendation" systems. Kendra Schaefer wrote out a detailed Twitter thread analyzing the proposal, and just how far reaching it is.

Among the new rules (as translated by by Stanford's DigiChina center) are some... um... interesting ones.

Algorithmic recommendation service providers shall uphold mainstream value orientations, optimize algorithmic recommendation service mechanisms, vigorously disseminate positive energy, and advance the use of algorithms upwards and in the direction of good.

Algorithmic recommendation service providers may not use algorithmic recommendation services to engage in activities harming national security, upsetting the economic order and social order, infringing the lawful rights and interests of other persons, and other such acts prohibited by laws and administrative regulations. They may not use algorithmic recommendation services to disseminate information prohibited by laws and administrative regulations.

It limits what kinds of keywords can be used. "Harmful" information cannot be used, nor "biased user tags" (whatever that means):

Algorithmic recommendation service providers shall strengthen user model and user tagging management and perfect norms for logging interests in user models. They may not enter unlawful or harmful information as keywords into user interests or make them into user tags to use them as a basis for recommending information content, and may not set up discriminatory or biased user tags.

There are also rules against using algorithms to manipulate systems -- but also to "shield information," which seems to contradict the requirements to block lots of information.

Algorithmic recommendation service providers may not use algorithms to falsely register users, illegally trade accounts, or manipulate user accounts; or for false likes, comments, reshares, web page navigation, etc.; or to carry out flow falsification or flow hijack. They may not use algorithms to shield information, over-recommend, manipulate topic lists or search result rankings, or control hot search terms or selections and other such interventions in information presentation; or to carry out self-preferencing, improper competition, influence on online public opinion, or evasion of supervision and management.

That one about "influencing public opinion" sure is interesting. How do you avoid influencing public opinion?

Like many recent regulatory proposals in the US, Canada, the EU and elsewhere, China leans heavily on "transparency" when it comes to algorithms:

Algorithmic recommendation service providers shall notify users in a clear manner about the situation of the algorithmic recommendation services they provide, and publicize the basic principles, purposes and motives, operational mechanisms, etc., of the algorithmic recommendation services in a suitable manner.

But, of course, viewing this in the Chinese context shows why such mandatory transparency has risks. In this case, China wants this transparency so it can further regulate what information people will see via algorithmic recommendations, and to intimidate/threaten companies not to spread information the government wants suppressed.

Again, similar to various internet regulations in the West, China has a nod towards end-user control and empowerment:

Algorithmic recommendation service providers shall provide users with a choice to not target their individual characteristics, or provide users with a convenient option to switch off algorithmic recommendation services. Where users choose to switch off algorithmic recommendation services, the algorithmic recommendation service provider shall immediately cease providing related services.

Algorithmic recommendation service providers shall provide users with functions to choose, revise, or delete user tags used for algorithmic recommendation services.

Where users believe algorithmic recommendation service providers use algorithms in a manner creating a major influence on their rights and interests, they have the right to require the algorithmic recommendation service provider to give an explanation and adopt related measures to improve or remedy the situation.

But, again, when viewed in the Chinese context, you can easily see how this kind of mandate can be abused heavily.

There is also, a "think of the children" provision, because no internet regulations these days are complete without such a moral panic heart string pull:

Where algorithmic recommendation service providers provide services to minors, they shall fulfill duties for the online protection of minors according to the law, and make it convenient for minors to obtain information content beneficial to their physical and mental health, through developing models suited for use with minors, providing services suited to the specific characteristics of minors, etc.

Algorithmic recommendation service providers may not push information content toward minor users that may incite the minor to imitate unsafe conduct, or acts violating social morals, or lead the minor towards harmful tendencies or may influence minors’ physical and mental health in other ways; and they may not use algorithmic recommendation services to lead minors to online addiction

Also, there's a built in complaint mechanism.

Algorithmic recommendation service providers shall accept social supervision, set up convenient complaints and reporting interfaces, and promptly accept and handle complaints and reports from the public.

Algorithmic recommendation service providers shall establish user appeals channels and mechanisms, to standardize the handling of user appeals and the timely provision of feedback, and realistically ensure the lawful rights and interests of users.

Failing to abide by these rules will get the companies fined (relatively small amounts -- about $1k to $5k -- at first). However, it also opens them up to significant criminal liability as well.

As Schaefer notes, this is China "going beyond" the internet regulations in the EU. But, what's left unsaid is that some of this is enabled by just how far the EU and others have gone in trying to get the internet to paper over the societal problems created by government failures in other realms.

It should be leading us to wonder why it is that China is so eager to embrace and extend this approach to internet regulations, rather than as an endorsement of such an approach.

Filed Under: algorithms, china, intermediary liability, internet regulations, online harms, online safety

Academic: Problems Created By Undermining Section 230 Can Be Solved... By Undermining Section 230?

from the say-that-again? dept

I remain confused at why so many people endorse Macarthur Genius award winner, Prof. Danielle Citron's views on Section 230. Over and over again people say that her ideas for reforming Section 230 are sensible. Except that they are not. She has falsely insisted that companies have no incentives to moderate and that their incentives are to push the most extreme content. This has been debunked over and over again. If it were true, then every website would turn into 8kun. But that doesn't happen, because most websites realize that when your website is full of garbage people, it drives away other users (including those more likely to support you or your advertisers) and it drives away advertisers.

Citron's big idea is to put in place a duty of care or "reasonableness" standard, but we've explained at length why that's the kind of idea only an academic with no experience running a website could love. It would lead to a ton of costly litigation in which companies would have to repeatedly defend their moderation practices in court. At best this would lead to companies all adopting nearly identical moderation practices to whichever company survived the litigation gauntlet first -- effectively crushing any ability to experiment and innovate in the moderation market, and likely locking in Facebook as one of the few companies that can afford to handle moderation's liability risks.

All that said, I'm even more flabbergasted by Citron's initial response to the (now walked back) news that OnlyFans was planning to phase out sexually explicit content. Many, many people have made the connection between OnlyFans' decision and FOSTA, the last time Section 230 was amended to add more liability to platforms. Of course, as we've now seen, FOSTA has had a massive human cost, is leading to some wacky vexatious civil lawsuits, and according to the GAO's own study, has failed to live up to any of its promises.

In an article at CNN about the OnlyFans decision, Citron makes an absolutely bizarre claim, that the answer to OnlyFans deplatforming sex workers... is to put more liability on OnlyFans so that those sex workers can sue. I only wish I were kidding.

Just like social media companies, payment processors are protected by Section 230 of the Communications Act of 1934, Citron said. That's the signature law that grants broad legal immunity to Facebook and Twitter for many of the content-moderation decisions they make — and the law that SESTA-FOSTA amended to create an exception for sex ads.

Citron wants to see changes made to Section 230 that could expose platforms to more liability under certain circumstances. Perhaps, she said, those changes might even allow sex workers who feel their businesses have been harmed by payment processors to sue them for tortious interference.

"We're talking about OnlyFans, where we're seeing sex workers doing safe work. It's from their own homes, they're making content on their own terms," Citron said.

So... the answer to sex workers being kicked off platforms that fear that they might face liability under FOSTA hole in Section 230, is to open up another hole in Section 230 so that those sex workers can then sue OnlyFans for tortious interference?!? Um, what?

Even just bringing up tortious interference as a cause of action is bizarre. That's the same sort of argument that people like Laura Loomer have used to say people reporting her to Twitter engaged in "tortious interference" when Twitter kicked her off the platform. Tortious interference in the context discussed here seems like a plan for vexatious nuisance litigation and SLAPP suits, not any kind of legitimate cause of action.

But, even more to the point, how is it a solution to the consequences of adding more liability to say "let's just pile on more liability to deal with those consequences"? It's liability all the way down. But it's completely disconnected from how all of this works in the real world.

Filed Under: danielle citron, duty of care, intermediary liability, liability, reasonableness, section 230
Companies: onlyfans

Last Month In An LA Court I Witnessed The Future Of A World Without Section 230; It Was A Mess

from the couldn't-you-have-done-more dept

Disclosure: I was an expert witness for RedBubble in this case (and another case) and submitted some reports regarding this case much earlier in the process, though I did not testify at this trial. I had nothing to do with this trial other than attending it as an interested observer, concerned about litigation involving content moderation. As you can see, however, my opinions on this remain identical to my opinions on content moderation going back basically forever...

There's been a lot of talk lately about "repealing" or "reforming" Section 230. We've explained at great lengthy why this would be a mistake and would lead to a ton of frivolous litigation. Section 230 co-author Senator Ron Wyden has referred to this as creating a situation where website operators would face "death by ten thousand duck-bites." Professor Eric Goldman has explained the procedural benefits of Section 230 in that it helps get rid of these cases more quickly. And while some (including myself) have pointed out that the 1st Amendment would still protect most claims, without the procedural benefits of Section 230, we'd still see a ridiculous number of nuisance lawsuits in which companies would have to defend each and every content moderation decision. This would be made significantly worse if the law changes in a manner proposed by folks like law professor Danielle Citron, to require websites to show that their content moderation practices are "reasonable," which would require expensive litigation to determine.

The thing is: we have an area where we can actually see this in practice. There is something of a loophole within Section 230: trademark law. From the beginning, Section 230(e)(2) has excluded intellectual property law. For copyright, that gap was filled (not particularly well) by the notice-and-takedown provisions of the DMCA Section 512. But trademark has no official intermediary protections in US law. Experts like Mark Lemley have long advocated for harmonizing the various intermediary laws so that they're all consistent. Many people had thought that perhaps the courts had settled the issue with regards to trademark in the infamous Tiffany case against eBay, in which the courts said that eBay shouldn't be expected to be perfect, and as long as it has some sort of program to deal with counterfeits, then it is no longer liable for counterfeits on its site.

However, last summer, a federal judge in LA effectively re-opened the issue in a case brought by the holding company that controls the "Brandy Melville" clothing brand against print-on-demand website RedBubble. Brandy Melville argued that RedBubble was violating its trademarks (and copyrights) in printing some of its designs, and while the judge did toss out much of the case, he allowed a few parts to move forward to a jury trial. The judge said that the Tiffany v. eBay rule didn't necessarily apply -- since that was a case of users selling physical products, whereas with RedBubble, the company would take designs uploaded by users, and pass them on to various print shops for the actual printing. So the open question, in theory, was whether or not that deeper involvement opened itself up to contributory liability (the judge correctly dismissed the claims about direct infringement). There were a few other oddities about the judge's ruling as Eric Goldman noted at the time:

However, the Ohio State ruling didn’t address contributory trademark infringement, and this court sends that issue to the jury. Unlike Tiffany v. eBay, which involved legitimate used resales of Tiffany items, the court says “Redbubble is not burdened by the need to parse valid uses of Brandy Melville’s trademarks from invalid ones, as Brandy Melville has made clear to Redbubble that there are no legitimate sales of Brandy Melville products through its site. As such, all uses of Brandy Melville’s marks that appear on Redbubble’s website are presumptively infringing.” I doubt the court means this statement so baldly, as it disregards parodies, nominative uses, and other legitimate activities. Similarly, the plaintiff complained that RedBubble hasn’t disabled internal consumer keyword searches for its trademarks, but that doesn’t seem like a right exclusive to the trademark owner in light of MTM v. Amazon.

In practice, however, pretty much the entire trial revolved around whether or not RedBubble "did enough" in its content moderation. Over and over again, the lawyers for Brandy Melville kept trying to insist that any failures by RedBubble's content moderation was proof that it did not do enough. Indeed, in the closing arguments, even after it had been pointed out that RedBubble spent millions of dollars per year on its content moderation efforts, and had removed every example that Brandy Melville's lawyers had sent them within a day or two, the lawyer straight up asked the jury "don't you think RedBubble could have done more?"

There were a number of other oddities that came out during the trial. The trial was explicitly about just three specific trademark designs that Brandy Melville held -- two of which never actually appear on the front of t-shirts or other products (the company said it only used them on clothing hang tags and on store signs), and one that the company registered just weeks before suing. Incredibly, Brandy Melville put on the stand a manager of some of their stores who admitted two incredible things: (1) that she was unable to tell which products on RedBubble were legitimate and which were infringing (so how the hell could they expect RedBubble to just know?!?) and (2) that in many years of working and managing Brandy Melville stores, she could not recall a single time that anyone ever brought in a RedBubble product to "return" it. That latter point really stuck with me because it highlighted the near total lack of confusion, which is a key point in trademark infringement. There needs to be confusion.

Another oddity: Brandy Melville kept focusing on the fact that RedBubble refused to block all tags that included "Brandy Melville" or any search results on "Brandy Melville." But as Goldman noted above (and the judge seemed to miss), that's not a right that trademark law gives a trademark holder. RedBubble pointed out that it actually used the tagging system as part of its content moderation efforts -- and that letting users tag stuff "Brandy Melville" made it easier for its content moderation team to find and review those items to see if they infringed. Also, while it only came up exceptionally briefly, one of the products that was on RedBubble's site was a t-shirt that read: "Fuck Brandy Melville," which is clearly commentary/parody -- and not something anyone would assume came from Brandy Melville itself. Brandy Melville cannot abuse trademark law to say any and all use of the term must be blocked. That's not how trademark law works.

For its part, RedBubble spent plenty of time describing the extreme lengths it goes to try to prevent infringing works on its site, including making use of various technologies, having a large team of human reviewers, and working closely with lots of brands to better understand what is and what is not infringing. But, again, the plaintiff's focused on "is it enough."

And, because they kept focusing on a very few examples where Brandy Melville logos or other designs were found on RedBubble, that was proof that, no, the company did not do enough. And... the jury bought it. At the end of the three day trial, the jury said that RedBubble was liable as a contributory infringer, including not just for trademark infringing, but for counterfeiting. This was particularly bizarre because for there to be counterfeiting, there needs to be an actual copying of a product sold by the trademark holder. And, as already noted, two out of the three designs were never actually used on the front of products by Brandy Melville, and as for the third design (the "LA Lightning" mark -- which, seems like it may have its own trademark issues, considering it was yanked from an independent professional basketball team), they never seemed to show exact copies of it on any RedBubble products either. Instead, somewhat bizarrely, Brandy Melville focused on very distinct and different variations on the logo that appeared on RedBubble products -- including a clear parody one for University of Florida students. And the counterfeiting issue was key to Brandy Melville's case, because it would allow them to ask for statutory damages, running up into multiple millions of dollars.

And this is a clear issue with litigation around content moderation: the jury was primed over and over again to believe that any mistake makes the company liable. Brandy Melville's lawyers just kept showing designs that mention Brandy Melville or looked similar to Brandy Melville (frequently for products that Brandy Melville had no trademark over) and then just kept harping on the fact that Brandy Melville didn't approve any of this, and clearly RedBubble was a "counterfeiting" site. In the end, the jury awarded $520,000 to Brandy Melville ($300k "counterfeiting" for the "Brandy Melville" mark, $200k for "counterfeiting" the LA Lightning mark, and then $20k for trademark infringement.)

At the close of the case, RedBubble immediately sought a judgment as a matter of law to effectively try to take the case out of the jury's hands (arguing that Brandy Melville failed to present enough evidence and the judge could rule on the matter directly without needing the factual deliberation of the jury). This week the judge did in fact rule that the $300k for contributory counterfeiting on the Brandy Melville mark was improper, bouncing the award down to $220,000.

As the judge noted, the fact that RedBubble's products were not the same as anything Brandy Melville made regarding those logos, meant that there couldn't be a legitimate counterfeiting claim for that mark:

Here, Plaintiff failed to present evidence of products that bore a spurious Brandy Melville Heart Mark and were offered for sale on Redbubble.com that were remotely similar to products that Plaintiff offered for sale, let alone "stich-for-stitch copies" of Plaintiff's products. The Court therefore concludes that the evidence presented at trial permits only one reasonable conclusion--Plaintiff failed to establish that Redbubble is liable for contributory counterfeiting of the Brandy Melville Heart Mark.

The end result is way below the many millions of dollars Brandy Melville was seeking, though even the small amount it got seems ridiculous based on an understanding of what actually happened.

But, there's a larger point here: this case was an unfortunate alternate reality universe view of what will happen without Section 230 or if we moved to a Citron-style "reasonable" standard. You'd get a ton of litigation in which everyone was just going back and forth arguing over what the "right" level of content moderation is, and arguing that any mistakes were evidence that "not enough" was being done by the companies. That's a disaster scenario, one in which any simple mistake (even where there's perhaps strong evidence that no mistake was actually made) is subject to the whims of a jury (or a judge) and an expensive, time-consuming trial. I recognize that law professors might not care that much about all of this wasteful litigation, but in the real world it matters a ton. This kind of thing can, and will, kill off smaller companies, even if they can succeed in court. And that doesn't help anyone, except possibly some trial lawyers who can threaten and shake down tons of websites.

Filed Under: content moderation, counterfeiting, intermediary liability, jury trials, print on demand, section 230, trademark
Companies: brandy melville, redbubble

Ninth Circuit Appeals Court Says Some Disturbing Stuff About Section 230 While Dumping Another Two 'Sue Twitter For Terrorism' Lawsuits

from the not-a-lot-of-good-news-in-the-167-pages dept

In a very dense and somewhat counterintuitive opinion [PDF], the Ninth Circuit Court of Appeals has dumped two more of the dozens of bogus "sue social media companies for acts committed by terrorists" lawsuits. But it has kept one alive. Worse, the 167-page ruling comes with concurring opinions that suggest Ninth Circuit judges think Section 230 immunity is, on the whole, letting social media companies get away with too much bad stuff.

The two lawsuits whose dismissals were affirmed deal with the San Bernardino shooting and the terrorist attacks in Paris, France. The one kept alive deals with a terrorist attack in Istanbul, Turkey. The first case (Gonzalez v. Google) deals with allegations that Google's revenue sharing with terrorist organizations amounted to material support. But the allegations aren't strong enough to sustain the lawsuit.

Although monetary support is undoubtedly important to ISIS’s terrorism campaign, the TAC is devoid of any allegations about how much assistance Google provided. As such, it does not allow the conclusion that Google’s assistance was substantial. Nor do the allegations in the TAC suggest that Google intended to assist ISIS. Accordingly, we conclude the Gonzalez Plaintiffs failed to state a claim for aiding-and-abetting liability under the ATA. We do not consider whether the identified defects in the Gonzalez Plaintiffs’ revenue-sharing claims—principally, the absence of any allegation regarding the amount of the shared revenue—could be cured by further amendment because the Gonzalez Plaintiffs were given leave to amend those claims and declined to do so.

The third case (Clayborn v. Twitter) involves several of the same allegations, but dealing with the San Bernardino shooting. The claims here fail mainly because there was no evidence ISIS directed or was directly involved in the attack.

Even if Congress intended “authorized” to include acts ratified by terrorist organizations after the fact, ISIS’s statement after the San Bernardino Attack fell short of ratification. The complaint alleges that ISIS stated, “Two followers of Islamic State attacked several days ago a center in San Bernardino in California, we pray to God to accept them as Martyrs.” This clearly alleges that ISIS found the San Bernardino Attack praiseworthy, but not that ISIS adopted Farook’s and Malik’s actions as its own.

The second case (Taamneh v. Twitter) stays alive. The allegations are pretty much identical to those in Gonzalez v. Google, with the main difference being what issues the lower court reached. In this case, there was no Section 230 discussion. And, as the Ninth Circuit sees it, that difference allows the claims it dismissed in Gonzalez to survive in Taamneh.

Because the bulk of the Gonzalez Plaintiffs’ claims were properly dismissed on the basis of § 230 immunity, our decision in Gonzalez principally focuses on whether the Gonzalez Plaintiffs’ revenue-sharing theory sufficed to state a claim under the ATA. In contrast, the district court in Taamneh did not reach § 230; it only addressed whether the Taamneh Plaintiffs plausibly alleged violations of the ATA for purposes of Rule 12(b)(6). The Taamneh appeal is further limited by the fact that the Taamneh Plaintiffs only appealed the dismissal of their aiding-and-abetting claim.

Hesitantly, the Ninth says this single lawsuit can proceed, mainly because it specified more direct support like use of Google's AdSense and other revenue-sharing.

We also recognize the need for caution in imputing aiding-and-abetting liability in the context of an arms-length transactional relationship of the sort defendants have with users of their platforms. Not every transaction with a designated terrorist organization will sufficiently state a claim for aiding-and-abetting liability under the ATA. But given the facts alleged here, we conclude the Taamneh Plaintiffs adequately state a claim for aiding-and-abetting liability.

Now, here comes the bad stuff. The concurring opinions don't deal much with the facts of the case, but rather with some judges' view that Section 230 is too broad and should be trimmed back. If Congress won't do it, maybe the Ninth Circuit will.

Here's Judge Marsha Berzon's take:

I concur in the majority opinion in full. I write separately to explain that, although we are bound by Ninth Circuit precedent compelling the outcome in this case, I join the growing chorus of voices calling for a more limited reading of the scope of section 230 immunity. For the reasons compellingly given by Judge Katzmann in his partial dissent in Force v. Facebook, 934 F.3d 53 (2d Cir. 2019), cert. denied, 140 S. Ct. 2761 (2020), if not bound by Circuit precedent I would hold that the term “publisher” under section 230 reaches only traditional activities of publication and distribution—such as deciding whether to publish, withdraw, or alter content—and does not include activities that promote or recommend content or connect content users to each other. I urge this Court to reconsider our precedent en banc to the extent that it holds that section 230 extends to the use of machine-learning algorithms to recommend content and connections to users.

The judge has problems with the recommendation algorithms used by social media companies -- ones that naturally tend to show people the sorts of things they appear to be interested in. In most cases, this is innocuous. But in some cases, the algorithms can send people down rabbit holes.

If viewers start down a path of watching videos that the algorithms link to interest in terrorist content, their immersive universe can easily become one filled with ISIS propaganda and recruitment. Even if the algorithm is based on content-neutral factors, such as recommending videos most likely to keep the targeted viewers watching longer, the platform’s recommendations of what to watch send a message to the user. And that message—“you may be interested in watching these videos or connecting to these people”—can radicalize users into extremist behavior and contribute to deadly terrorist attacks like these.

This is a really weird, really dangerous place to start drawing the line in Section 230 lawsuits. Algorithms react to input from users. If YouTube can't be held directly responsible for videos uploaded by users, it makes sense it would be immunized against algorithmically suggesting content based on users' actions and preferences. The algorithm does next to nothing on its own without input from content viewers. It takes a user to really get it moving.

Judge Ronald Gould's concurrence contains many of the same complaints about social media recommendation algorithms. And he similarly believes Section 230 shouldn't cover these, apparently for the simple reason that terrorists can benefit from recommendations made to users who've expressed an affinity for content allegedly created by terrorists.

The majority ultimately concludes that Section 230 shields Google from liability for its content-generating algorithms. I disagree. I would hold that Plaintiffs’ claims do not fall within the ambit of Section 230 because Plaintiffs do not seek to treat Google as a publisher or speaker of the ISIS video propaganda, and the same is true as to the content-generating methods and devices of Facebook and Twitter.

Accepting plausible complaint allegations as true, as we must, Google, through YouTube, and Facebook and Twitter through their various platforms and programs, acted affirmatively to amplify and direct ISIS content, repeatedly putting it in the eyes and ears of persons who were susceptible to acting upon it.

And if Congress won't act fast enough for Judge Gould, then the courts should step in and regulate social media companies.

I further urge that regulation of social media companies would best be handled by the political branches of our government, the Congress and the Executive Branch, but that in the case of sustained inaction by them, the federal courts are able to provide a forum responding to injustices that need to be addressed by our justice system. Here, that means to me that the courts should be able to assess whether certain procedures and methods of the social media companies have created an unreasonably dangerous social media product that proximately caused damages, and here, the death of many.

Judge Gould says this should be easy to do correctly and without collateral damage to legitimate content.

The record shows that despite extensive media coverage, legal warnings, and congressional hearings, social media companies continued to provide a platform and communication services to ISIS before the Paris attacks, and these resources and services went heedlessly to ISIS and its affiliates, as the social media companies refused to actively identify ISIS YouTube accounts, and only reviewed accounts reported by other YouTube users. If, for example, a social media company must take down within a reasonable time sites identified as infringing copyrights, it follows with stronger logic that social media companies should take down propaganda sites of ISIS, once identified, within a reasonable time to avoid death and destruction to the public, which may be victimized by ISIS supporters. Moreover, if social media companies can ban certain speakers who flout their rules by conveying lies or inciting violence, as was widely reported in the aftermath of tweets and posts relating to the recent “insurrection” of January 6, 2021, then it is hard to see why such companies could not police and prohibit the transmission of violent ISIS propaganda videos, in the periods preceding a terrorist attack.

This ignores the fact that the DMCA process is pretty much an ongoing train wreck, one that's abused to silence speech and often mistakenly targets non-infringing content. And social media companies' attempts to stop the spread of disinformation or deal with harassing/threatening content have rarely been viewed as competent, much less exemplary. This whole spiel ignores the fact that a lot of the moderation Gould considers easy or successful is also heavily reliant on reports by site users.

Finally, not only does Judge Gould suggest Section 230 should be narrowed, he thinks another course of legal action should be made available to plaintiffs to sue tech companies not just for the content they host, but for the actions of terrorist organizations all over the world.

As a matter of federal common law, I would hold that when social media companies in their platforms use systems or procedures that are unreasonably dangerous to the public—as in the case where their systems line up repeated messages in aid of terrorists like ISIS—or when they omit to act to avoid harm when omitting the act is unreasonably dangerous to the public—as in the case where they fail to review and self-regulate their websites adequately to notice and remove propaganda videos from ISIS that are likely to cause harm—then there should be a federal common law claim available against them.

That's Gould's "product liability" theory. In his view, the algorithms are defective because some users turn towards terrorism. And if the product is defective, the manufacturer can be sued. Gould really has to stretch the analogy to make it fit.

Here and similarly, social media companies should be viewed as making and “selling” their social media products through the device of forced advertising under the eyes of users.

Huh. But what if users use ad blockers? I mean, that's just one of several questions this raises. The product is access to site users and their attention. That's what's being sold to advertisers. If that's the case, the only parties who would have standing to bring lawsuits under this theory would be dissatisfied companies who feel their ads are being placed along content that's being served up by algorithms that are possibly radicalizing some users into committing terrorist acts. That's more than a little attenuated from the actual harm, especially if no one working for the aggrieved companies has been a victim of a terrorist attack. Sure, users could try to complain the product is defective, but the product -- according this this judge's own take -- isn't the social media platform.

Look, moderation is far from perfect and will likely always cause some sort of collateral damage as adjustments are made. If everyone would like to see less moderation and fewer social media options, they should definitely allow the courts and Congress to start creating a bunch of exceptions to Section 230 immunity. In these three lawsuits, plaintiffs suffered tragedies and were encouraged by questionable law firms to sue third parties with no link to terrorists and their acts other than some hosted content. If these claims had any merit, we'd see more wins. But we haven't seen these wins because these claims are weak and seem mostly propelled by finding the largest, easiest target to hit rather than any true desire to see justice done.

Filed Under: 9th circuit, ata, intermediary liability, section 230, terrorism
Companies: twitter

Appeals Court Tosses Cop's Attempt To Hold Twitter Responsible For Him Being Shot By A Gunman

from the law-firms-must-be-hoping-for-a-first-round-draft-pick dept

The 1-800-LAW-FIRM/Excolo Law losing streak continues. The lawyers at those firms have been preying on the victims of terrorist attacks for a few years now, presumably promising them some form of justice that can only be obtained by filing futile, flawed, completely doomed lawsuits that attempt to hold social media companies directly responsible for the criminal acts of terrorists.

This lawsuit -- facing its second rejection in a row -- was one of the stupidest filed by these law firms. Jesus Retana, a Dallas police officer, was wounded during a mass shooting perpetrated by Micah Johnson. The lawsuit attempted to use the AITA (Anti-International Terrorism Act) to hold Twitter responsible for the alleged "radicalization" of Johnson by Hamas, an entity the US government has declared a terrorist organization.

The Fifth Circuit Appeals Court finds this attempt to tie Twitter to Micah Johnson and his shooting of Dallas police officers as legally unbelievable as the lower court did. There's no discussion of Section 230 immunity, which would shield Twitter from this litigation, because the rest of the allegations are so baseless there's no reason to bring it up. (h/t Eric Goldman)

The court says [PDF] not only is there no link between Hamas, Twitter, and Micah Johnson, there's also no link between Hamas and the shooting in Dallas. And there's certainly no international terrorism justifying the accusations of Twitter's supposed violation of international terrorism laws.

An act of international terrorism is one that “occur[s] primarily outside the territorial jurisdiction of the United States, or transcend[s] national boundaries.” We cannot conclude that the Dallas shooting transcended national boundaries. We sympathize with Plaintiffs, but this shooting was committed by a lone shooter entirely within the United States. He might have been radicalized in part by Hamas, but Hamas did not plan the shooting or even take credit for it.

There's also no connection between Johnson and Hamas in the 96 pages of the original lawsuit, nor in its current amended state. Linking Johnson to Hamas is crucial, especially if the plaintiffs want to claim Twitter helped Hamas radicalize the shooter. That link simply isn't there.

Johnson was a self-radicalized shooter who merely “liked” the Facebook pages of “black separatist hate groups” that had communicated with Hamas. And, while Johnson met with Ms. X and indicated that he sympathized with Hamas’s objectives, that occurred two years prior to the Dallas shooting.

It was a shooting perpetrated by a US citizen on US soil with no solid ties to any international terrorist group. No international terrorist group took credit for the attack. There was no international terrorism. The AITA claim fails. Again.

So does the allegation of secondary liability.

We also hold that Defendants did not “knowingly” and “substantial[ly]” assist Hamas in the Dallas shooting, again because the shooting was committed by Johnson alone and not by Hamas either alone or in conjunction with Johnson. The district court was correct in concluding that Defendants are not secondarily liable under the ATA.

This case was dismissed by the district court in December 2019. At that point, it was these law firms' TWELFTH straight loss in federal courts. That win-loss record hasn't improved since then. Unfortunately, these firms seem to have no problem finding new victims of terrorist acts to use as fodder for DOA lawsuits.

Filed Under: blame, intermediary liability, jesus retana, micha johnson, responsibility, terrorism
Companies: 1-800-lawfirm, excolo law, twitter

As Everyone Rushes To Change Section 230, New GAO Report Points Out That FOSTA Hasn't Lived Up To Any Of Its Promises

from the garbage-in,-garbage-out dept

As you may have heard, tons of politicians are rushing to introduce new and different bills to undermine or repeal Section 230 of the Communications Decency Act -- a bill that is rightly credited for enabling a more open internet for freedom of speech. As you may recall, in early 2018 we had the first actual reform to Section 230 in decades -- FOSTA. It was signed into law on April 11th, with tons of politicians insisting it was critical to protecting people online. We had so many quotes from politicians (and a whole campaign from Hollywood stars like Amy Schumer) claiming (falsely) that without FOSTA, children could be "bought and sold" online.

One thing the bill did include (in Section 8) was a requirement that 3 years after the bill passed, the GAO should put out a report on how effective it has been. It's a few months late (the GAO does excellent work, but tends to be overworked and under-resourced) but on Monday the GAO finally released its study on the effectiveness of FOSTA. And... it basically says that all of the critics claims were exactly right.

Before FOSTA became law, co-author of Section 230, Senator Ron Wyden warned:

I fear that the legislation before the Senate will be another failure. I fear it will do more to take down ads than take down traffickers. I fear it will send the bad guys beyond the grasp of law enforcement to the shadowy corners of the dark web, where everyday search engines don’t go, but where criminals find safe haven for their monstrous acts....

[....]

In my view, the legislation before the Senate will prove to be ineffective, it will have harmful unintended consequences, and it could be ruled unconstitutional.

[....]

But the bill before us today will not stop sex trafficking. It will not prevent young people from becoming victims.....

So, now that we're three years in, what has the GAO found? That, just as predicted, the law was not at all necessary, and has barely been used, and the very, very few times it's been used in court, it's been ineffective:

Criminal restitution has not been sought and civil damages have not been awarded under section 3 of FOSTA. In June 2020, DOJ brought one case under the criminal provision established by section 3 of FOSTA for aggravated violations involving the promotion of prostitution of five or more people or acting in reckless disregard of sex trafficking. As of March 2021, restitution had not been sought or awarded. According to DOJ officials, prosecutors have not brought more cases with charges under section 3 of FOSTA because the law is relatively new and prosecutors have had success using other criminal statutes. Finally, in November 2020 one individual sought civil damages under a number of constitutional and statutory provisions, including section 3 of FOSTA. However, in March 2021, the court dismissed the case without awarding damages after it had granted defendants' motions to dismiss.

So... why did we need it again? Why was it so urgent? Why were Senators, Members of Congress, Hollywood stars, and others practically shoving each other aside to say that we needed this yesterday? And now, it's barely been used?

The report also shows -- again as many of us predicted -- that post FOSTA, law enforcement has had more difficulty tracking down those engaged in sex trafficking. Not because there is less trafficking, but because they're harder for law enforcement to find or access the details:

The current landscape of the online commercial sex market heightens already-existing challenges law enforcement face in gathering tips and evidence. Specifically, gathering tips and evidence to investigate and prosecute those who control or use online platforms has become more difficult due to the relocation of platforms overseas; platforms’ use of complex payment systems; and the increased use of social media, dating, hookup, and messaging/communication platforms.

The relocation of platforms overseas makes it more difficult for law enforcement to gather tips and evidence. According to DOJ officials, successfully prosecuting those who control online platforms—whether their platforms are located domestically or abroad—requires gathering enough evidence to prove that they intended that their platforms be used to promote prostitution, and, in some cases, that they also acted in reckless disregard of the fact that their actions contributed to sex trafficking.

Of course, in the runup to passing FOSTA, everyone kept talking about Backpage -- that FOSTA was needed to takedown Backpage, the company that everyone insisted was terrible. Except... as everyone now knows, the DOJ took down Backpage without FOSTA. Perhaps, the real issue had nothing to do with the law itself, and plenty to do with the DOJ not doing much on this issue. Or, worse, the DOJ recognizing that maybe Backpage wasn't the evil bogeyman that politicians and the press were making it out to be.

After all, leaked government documents later showed that Backpage was actually working closely with law enforcement to track down traffickers. So, guess what the report has found now? Thanks to FOSTA and also to the takedown of Backpage, the FBI is now finding it really really difficult to actually track down traffickers:

According to a 2019 FBI document, the FBI’s ability to identify and locate sex trafficking victims and perpetrators was significantly decreased following the takedown of backpage.com. According to FBI officials, this is largely because law enforcement was familiar with backpage.com, and backpage.com was generally responsive to legal requests for information. In contrast, officials said, law enforcement may be less familiar with platforms located overseas. Further, obtaining evidence from entities overseas may be more cumbersome and time-intensive, as those who control such platforms may not voluntarily respond to legal process, and mutual legal assistance requests may take months, if not years, according to DOJ officials.

So... end result: the government is barely using FOSTA and it's now significantly more difficult to find sex traffickers.

And, of course, the report doesn't even touch on the fact that things FOSTA did do to harm sex workers, putting them more at risk. It also doesn't talk about how lots of legitimate sites, such as dating sites and Tumblr, started aggressively blocking content that was likely perfectly legal, out of fear that FOSTA would open them up to criminal liability.

So, here's the big question: as a ton of politicians are pushing for big massive changes to Section 230, will they listen to us this time when we warn them about the possible consequences of such changes? Or will they dismiss us and insist that we're lying like they did last time? And who will go ask the politicians and Hollywood stars who swore that FOSTA was so absolutely necessary, how they respond to the fact that it didn't work, isn't being used, has made it more difficult to stop actual trafficking and has put actual lives in danger? Because all of that seems kind of important.

Filed Under: fosta, gao, intermediary liability, law enforcement, section 230, sex trafficking

Why The Ninth Circuit's Decision In Lemmon V. Snap Is Wrong On Section 230 And Bad For Online Speech

from the another-hard-case dept

Foes of Section 230 are always happy to see a case where a court denies a platform its protection. What's alarming about Lemmon v. Snap is how comfortable so many of the statute's frequent defenders seem to be with the Ninth Circuit overruling the district court to deny Snapchat this defense. They mistakenly believe that this case raises a form of liability Section 230 was never intended to reach. On the contrary: the entire theory of the case is predicated on the idea that Snapchat let people talk about something they were doing. This expressive conduct is at the heart of what Section 230 was intended to protect, and denying the statute's protection here invites exactly the sort of harm to expression that the law was passed to prevent.

The trouble with this case, like so many other cases with horrible facts, is that it can be hard for courts to see that bigger picture. As we wrote in an amicus brief in the Armslist case, which was another case involving Section 230 with nightmarish facts obscuring the important speech issues in play:

"Tragic events like the one at the heart of this case can often challenge the proper adjudication of litigation brought against Internet platforms. Justice would seem to call for a remedy, and if it appears that some twenty-year old federal statute is all that stands between a worthy plaintiff and a remedy, it can be tempting for courts to ignore it in order to find a way to grant that relief."

Here some teenagers were killed in a horrific high-speed car crash, and of course the tragedy of the situation creates an enormous temptation to find someone to blame. But while we can be sympathetic to the court's instinct, we can't suborn the facile reasoning it employed to look past the speech issues in play because acknowledging them would have interfered with the conclusion the court was determined to reach. Especially because at one point it even recognized that this was a case about user speech, before continuing on with an analysis that ignored its import:

Shortly before the crash, Landen opened Snapchat, a smartphone application, to document how fast the boys were going. [p.5] (emphasis added)

This sentence, noting that the boys were trying to document how fast they were going, captures the crux of the case: that the users were using the service to express themselves, albeit in a way that was harmful. But that's what Section 230 is built for, to insulate service providers from liability when people use their services to express themselves in harmful ways because, let's face it, people do it all the time. The court here wants us to believe that this case is somehow different from the sort of matter where Section 230 would apply and that this "negligent design" claim involves a sort of harm that Section 230 was never intended to apply to. Unfortunately it's not a view supported by the statutory text or the majority of precedent, and for good reason because, as explained below, it would eviscerate Section 230's critical protection for everyone.

Like it had done in the Homeaway case, the court repeatedly tried to split an invisible hair to pretend it wasn't trying to impose liability arising out of the users' own speech. [See, e.g., p. 10, misapplying Barnes v. Yahoo]. Of course, a claim that there was a negligent design of a service for facilitating expression is inherently premised on the idea that there was a problem with the resulting expression. And just because the case was not about a specific form of legal liability manifest in their users' speech did not put it outside of Section 230. Section 230 is a purposefully broadly-stated law ("No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider."), and here the court wants the platform to take responsibility for how its users used its services to express themselves. [p. 15, misapplying the Roommates.com case].

Section 230 also covers everything that could be wrong with expression unless the thing wrong with it happens to fall into one of the few exceptions the statute enumerates: it involves an intellectual property right, violates federal criminal law, or otherwise implicates FOSTA. None of those exceptions apply here, and, in fact, in the same section of the law where these few exceptions are set forth there is also a pre-emption provision explicitly barring any state law from becoming the basis of any new exceptions. Which, with this decision giving the go-ahead to a state law-based tort claim of "negligent design," is what the Ninth Circuit has now caused to happen.

It hurts online speech if courts can carve out new exceptions. If judges can ever post hoc look at a situation where expressive activity has led to harm and decide the degree of harm warrants stripping service providers of their Section 230 protection, then there is basically no point in having Section 230 on the books. If platforms have to litigate over whether it protects them, then it doesn't really matter whether it does or not because they'll already have lost out on so much of the value the protection was supposed to afford them to make it possible for them to facilitate others' expression in the first place. The inevitable consequence of this functional loss of statutory protection is that there will be fewer service providers available to facilitate as much user expression, if any at all.

But even if there were some limiting principle that could be derived from this case to constrain courts from inventing any other new exceptions, just having this particular "negligent design" one will still harm plenty of speech. To begin with, one troubling aspect the decision is that it is not particularly coherent, and one area of confusion relates to what it actually thinks is the negligent design. [see, e.g., p. 15]. The court spends time complaining about how Snapchat somehow deliberately encourages users to drive at unsafe speeds, even though the court itself acknowledged that while Snapchat apparently rewards users with "trophies, streaks, and social recognitions" to encourage them to keep using their service [p. 5], it "does not tell its users how to earn these various achievements" [p. 5], and it is a leap to say that Snap is somehow wrongfully encouraging users to do anything when it is not actually saying anything of the kind. [See p. 6 ("Many of Snapchat’s users suspect, if not actually 'believe,' that Snapchat will reward them for 'recording a 100-MPH or faster [s]nap' using the Speed Filter.")]. In fact, as the decision itself cites, Snapchat actually cautioned against reckless posting behavior. [See p. 6 with the screenshot including the text, "Don't snap and drive."] If the case were actually about Snap explicitly encouraging dangerous behavior ("Drive 100 mph and win a prize!") then there might legitimately be a claim predicated on the platform's own harmful speech, for which Section 230 wouldn't apply. But the record does not support this sort of theory, the theory of liability was predicated on a user's apparently harmful speech, and in any case the alleged encouragement wasn't really what the plaintiffs were charging was actually negligently designed anyway.

Instead, what was at issue was the "speed filter," a tool that helped users document how fast they were traveling. Unlike the district court, the Ninth Circuit could not seem to fathom that a tool that helped document speed could be used for anything other than unsafe purposes. But of course it can. Whether traveling at speed is dangerous depends entirely on context. A user in a plane could easily document traveling at significant speed perfectly safely, while a user on a bike documenting travel at a much slower speed could still be in tremendous peril. One reason we have Section 230 is because it is impossible for the service provider to effectively police all the uses of its platform, and even if it could, it would be unlikely to know whether the speeding was safe or not. But in denying Snapchat Section 230 protection with the presumption that such speech is always unsafe, the court has effectively decided that no one can ever document that they are traveling quickly, even in a safe way, because it is now too legally risky for the platform to give users the tools to do it.

Furthermore, if a platform could lose its Section 230 platform because the design of its services enabled speech that was harmful, it would eviscerate Section 230, because there are few, if any, whose design would not. For example, Twitter's design lets people post harmful expression. Perhaps one might argue it even encourages them to by making it so easy to post such garbage. Of course, Twitter also makes it easy to post things that are not harmful too, but the Ninth Circuit's decision here does not seem to care that a design eliciting user expression might be used for both good and bad ends. Per this decision, which asserts a state law-created "duty to design a reasonably safe product," [see p. 13, misapplying the Doe 14 v. Internet Brands case], even a product that meets the definition of an "interactive computer service" set forth in Section 230 (along with its pre-emption provision), if the design could be used to induce bad expression, then the platform no longer qualifies for Section 230's protection. But that would effectively mean that everyone could always plead around Section 230 because nearly every Section 230 case arises from someone having used the service in a harmful way the service enabled. It is unfortunate that the Ninth Circuit has now opened the door to such litigation, as the consequences stand to be chilling to all kinds of online speech and services Section 230 was designed to protect.

Filed Under: 9th circuit, intermediary liability, lemmon, negligence, product liability, section 230, speech, speed filter
Companies: snap

Are Partial Liability Rules The Path Forward For Intermediary Liability Regimes? Lessons Learned From Brazil

from the marco-civil dept

The past year saw the Internet become a lifeline during the COVID-19 pandemic. But 2020 also saw increased scrutiny of online content moderation, regulation of platforms and their effects on society. While recent headlines have focused heavily on social media platforms, the conversation is much more complex: the future of the Internet as we know it depends on discussions and policies regarding intermediary liability -- the legal rule that platforms should not be liable for the content posted by their users.

Section 230 reform in the U.S. and the proposed DSA and DMA in the European Union are driving a new era of intermediary liability rule-making, and other countries have followed suit: India recently updated guidelines applicable to intermediaries, and Mexico is discussing how to ensure freedom of expression in social media platforms. Instead of letting the U.S. and Europe influence these debates, governments around the world can learn important lessons from Brazil.

With approximately 145 million Internet users, Brazil has a large and growing digital economy. There are around 10,000 ISPs operating in the country, broadband Internet is available to almost 90% of the population, and the country's Internet Exchange Point, IX.br, is one of the largest in the world. Brazil also has a strong tradition of Internet governance and policy. Since 1995, the country's "Multistakeholder Internet Steering Committee" (CGI.br) has provided technology and policy recommendations to stakeholder groups to leverage the full potential of the Internet. In 2014, the country adopted an "Internet Bill of Rights," which establishes rights and duties of individual and corporate users, businesses and the government.

Brazil has long been a pioneer in sound Internet policies and regulation and holds one of the most influential laws regarding intermediary liability not only in Latin America, but the world. The Brazilian "Marco Civil da Internet", or Civil Rights Framework for the Internet, which was approved in 2014, introduced an intermediary liability regime built upon almost two decades of practice and jurisprudence in the U.S. around CDA's Section 230.

However, Brazil decided to deviate from the path enshrined in the 1996 U.S. legal order. The Marco Civil law grants full immunity from liability to Internet access providers and clearly indicates that Internet infrastructure should not be affected by issues pertaining to the upper layers (i.e., applications layer) of the network. This law adopts a restricted and residual approach: Internet application providers are only held liable for third-party content in instances where they fail to comply with specific judicial orders to render certain content unavailable. As a general rule, private notice and take down notifications are not accepted as a means to trigger the liability system comprised in the Marco Civil. The rule however does not apply to non-consensual dissemination of intimate images and matters affecting intellectual property - the latter being subject to a specialized regime.

Perhaps the most important difference between Marco Civil and Section 230 is that Brazil has deliberately decided not to copy the "Good Samaritan" clause, meaning that the 'protection' provided by the Marco Civil in Article 19 does not grant any immunity to content moderation practices adopted by Internet application providers.

While the U.S. system encompasses ante-hoc immunity for liability from third-party content and also for a company's own good-faith behavior vis-à-vis its users, the Brazilian system covers only third-party content. In Brazil, there is no ante-hoc immunity whatsoever for harms caused by the decisions and measures taken by Internet application providers. For instance, Google was recently forced to pay compensation for "authentic ante-hoc censorship" when it applied its terms of service to remove videos from a Civil Society organization's YouTube channel. What in the U.S. would be solved by the application of the "Good Samaritan" clause, in Brazil had a different outcome.

Why should we care about Brazil's Marco Civil and why now? Around 90% of Internet users are outside the U.S. and the narrow application of the Section 230, as well as the full immunities it grants, might no longer work to guide Internet policy development elsewhere (especially considering the complex and diverse discussions related to the scope of Freedom of Speech). Also, the various proposed reforms to Section 230 in the U.S. have spilled over to other countries in very dangerous ways, including in Brazil. In their fight against Big Tech, some politicians in Brazil -emulating the behavior of US politicians- have proposed to suppress immunities that do not really exist in our legal order (as our overarching liability regime already covers issues such as the wrongful suppression of content by application providers). So, in addition to being a waste of time, these attempts are counterproductive as the country could take steps backward in Internet policy and regulation. .

The partial intermediary liability adopted by the Brazilian Marco Civil created obstacles for extrajudicial requests for content removal that throughout the 2000s helped foster legal uncertainty and very little transparency and accountability from platforms. By granting Internet applications with immunity solely for third-party behavior and content (and not for their own behavior and practices), Marco Civil contributed to increased legal predictability and fostered innovation in Brazil, as shown in by a study commissioned by the Internet Society.

The current landscape of platform and intermediary liability rulemaking marks an ethos change from previous decades. In the past, it was imperative to avoid regulating or harming the Internet through overregulation. Today regulation is no longer a taboo. However, regulation must be wise, principled-based and aimed at the correct target in the complex digital ecosystem.

Regulators must resist adopting policies that penalize the behavior of social media platforms as they ultimately punish all users who post content on these platforms. Rules aimed at social media platforms will end up impacting other types of applications for the mere fact that they operate on the same layer of the Internet. Any regulation that forces Internet infrastructure providers to enter the business of content moderation would be dangerous. The consequences for freedom of expression, innovation and a dynamic digital economy would be significant.

How governments decide to address intermediary liability in the near future is critical for users and for the Internet. We face a critical juncture where we can either get this right or get it wrong. Learning from Brazil's Marco Civil law is a step in the right direction.

Bruna Martins dos Santos is Advocacy Coordinator for Data Privacy Brazil Research and Diego Canabarro is Senior Policy Manager for the Internet Society.

Filed Under: brazil, intermediary liability, marco civil