Sections

Research

The politics of Section 230 reform: Learning from FOSTA’s mistakes

A top-down view of a person typing on a laptop in the dark.

In July 2021, President Biden lost his patience with Facebook. “They’re killing people,” the president said of the falsehoods around the coronavirus and COVID-19 vaccines circulating on the platform. A White House spokeswoman announced that the administration was “reviewing” Section 230 (§ 230) of the Communications Decency Act, the law that protects Facebook and other platforms from being held liable for much of what users post to their websites. “[C]ertainly they should be held accountable,” she told CNN.

Just three months later, as Facebook struggled to respond to allegations of recklessness and mismanagement by whistleblower Frances Haugen, the idea of tinkering with § 230 came up again. Responding to documents released by Haugen about the harms potentially caused by Facebook’s recommendation algorithms, a group of Democratic members of Congress introduced legislation to open up websites to lawsuits for certain content boosted by the platform. “Under this bill,” announced Rep. Mike Doyle of Pennsylvania, “Section 230 would no longer fully protect social media platforms from all responsibility for the harm they do to our society.”

For most of its 26-year existence, § 230 was familiar mostly just to lawyers and technologists. Now, though, it seems to be every American politician’s favorite scapegoat when it comes to technology policy, and the statute has spent enough time in the headlines that journalists and commentators have developed a habit of referring to it as “once-obscure.” In the 116th Congress, lawmakers formally introduced more than 25 bills to amend or repeal § 230. So far, the 117th Congress is keeping pace, with over 20 reform proposals introduced. The statute is one of the few issues on which President Biden and his predecessor agree: President Trump tweeted dozens of times for § 230 to be repealed, and Biden commented during the 2020 Democratic primary that the law should be done away with. The ongoing arguments over whether and how to regulate the internet pull in a number of different directions, but almost everyone seems to agree that something should be done about § 230—even if nobody can agree what that something is.

Yet, amid all this debate, there has been practically no discussion on Capitol Hill of the last time that Congress revised § 230—just three years ago. In 2018, President Trump signed into law the Allow States and Victims to Fight Online Sex Trafficking Act (FOSTA). Among other provisions, FOSTA carved an exception out of § 230’s liability shield for websites that “knowingly assist[], support[], or facilitate[]” activity violating federal sex trafficking law, opening internet companies up to a potential wave of lawsuits over material for which they’d previously received legal protection. The bill passed with overwhelming support in both chambers of Congress. In the Senate, only two lawmakers voted no: Sen. Ron Wyden (D-Ore.), who in 1996 was one of § 230’s original authors, and libertarian Sen. Ron Paul (R-Ky.).

FOSTA would seem to be an obvious touchstone for discussions around § 230 reform today. It’s not only a recent precedent, but one that enjoyed bipartisan support by a wide margin—no small feat in a polarized time. Three years after its passage, however, the law does not make for an encouraging model. Studies suggest that the law likely placed sex workers at increased risk, as people lost access to websites used to advertise and communicate about the sex trade and were forced to work on the street in dangerous conditions. And apart from FOSTA’s changes to § 230, a July 2021 report by the Government Accountability Office (GAO) showed that federal prosecutors have had little use for the additional criminal penalties for sex trafficking established by FOSTA. The law may even have hindered efforts to investigate and prosecute trafficking.

But Congress seems to have little interest in reviewing its past work. The GAO report arrived to little fanfare from Capitol Hill. And FOSTA remains almost entirely absent from current political discussions around amending § 230.

That’s unfortunate because, when it comes to § 230 reform, FOSTA deserves greater attention as a cautionary tale. The law is “a case study as to how platforms react to new § 230 exemptions,” writes Jeff Kosseff, an associate professor at the U.S. Naval Academy and the author of “The Twenty-Six Words that Created the Internet,” a book about § 230. In response to FOSTA’s broad and confusing exceptions to liability protection under § 230, a wide range of companies shut down or limited access to online services, including services that had little or nothing to do with sex trafficking. As the first major revision to § 230 since the statute became law over two decades ago, FOSTA’s ripple effects demonstrate why lawmakers should move with extreme care in attempting to revise § 230.

The two-decade-long journey from Section 230 to FOSTA is, in large part, the story of how American views of the internet soured.

The two-decade-long journey from § 230 to FOSTA is, in large part, the story of how American views of the internet soured. And yet, § 230 only rocketed to political prominence in the years after Congress passed FOSTA. Studying the cultural and political currents behind those transformations—the internet from symbol of a bright future to bleak omen, and § 230 from obscure to reviled—helps explain both how FOSTA is a part of that larger story and what is lost when the law is left out of the story’s telling.

From Section 230 to FOSTA

Section 230 became law in 1996, when the internet was in its infancy as a force in popular life. By 2017, when lawmakers first began drafting the legislation that would become FOSTA, that early shine had worn off—tarnished by, among other things, Russian election interference in the 2016 U.S. presidential election. The Economist memorably termed this change in mood the “techlash,” an abrupt cooling in public feelings toward major technology platforms after 2016.

Before the techlash, § 230 might have seemed an unlikely target for widespread public ire. But it has now become a shorthand for more generalized discontent over the state of the internet—even if the specific complaint in question is unrelated to the law itself.

For a statute that has caused so much confusion, the basic idea behind § 230 is simple. Generally speaking, the law shields websites from being held legally responsible for content that others post—a protection not available for print material or television broadcasts. If I post something defamatory about you on Twitter, for example, you can sue me, but you can’t sue Twitter. This is established by § 230(c)(1), which states that “[n]o 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.” The statute also contains a second major provision, § (c)(2), which states:

No provider or user of an interactive computer service shall be held liable on account of—

(A) any action voluntarily taken in good faith to restrict access to or availability of material that the provider or user considers to be obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable, whether or not such material is constitutionally protected; or

(B) any action taken to enable or make available to information content providers or others the technical means to restrict access to material described in paragraph (1).

In brief, as courts have interpreted the law, § 230 (c)(1) protects platforms from civil liability for leaving content up; § 230 (c)(2) protects them if they choose to take content down. While § 230 shields platforms from federal civil claims, the statute also protects against litigation “under any State or local law that is inconsistent with this section.” It includes further exemptions for federal criminal law, intellectual property law, and the Electronic Communications Privacy Act.

To understand the significance of § 230’s liability shield, consider the legal risk that platforms faced before the statute. In the 1991 case Cubby v. CompuServe, a federal district court ruled that the internet service provider CompuServe could not be held liable for allegedly defamatory posts published to its platform, on the grounds that CompuServe was passively hosting the material without moderating or reviewing it, like a bookstore stocking a wide range of volumes. Just like a bookstore, CompuServe couldn’t be expected to know the contents of every piece of content on its site.

Five years later, in New York state court, another judge turned this reasoning against Prodigy—a CompuServe competitor that advertised itself as a “family-oriented” computer service and touted its willingness to moderate content on its platform as a selling point. Prodigy had exerted control over what went up on its services, the court reasoned, and therefore was subject to liability not as a distributor but a publisher. Like a newspaper liable for everything published within its pages, Prodigy could be held legally responsible for what users posted on its site, even if the company were unaware of the particular post at issue.

This case, titled Stratton Oakmont v. Prodigy, created what Santa Clara University law professor Eric Goldman has called the “moderator’s dilemma.” Given that perfect content moderation is impossible and some problematic material will always slip through, is it better to attempt to moderate posts by users, knowing that any failures to remove illegal or unsavory material might open up platforms to legal liability? Or should platforms instead follow CompuServe’s example, maximizing their ability to plead ignorance by simply giving up moderation efforts altogether? Section 230 attempted to resolve this dilemma by removing the legal risks of moderation.

Soon after Stratton Oakmont was decided, Wyden (then a member of the House of Representatives) and then-Rep. Chris Cox (R-Calif.) began working on what would become § 230. The two drafted the provision in part as an alternative to legislation, introduced by then-Sen. James Exon (D-Neb.), that would have imposed criminal penalties for transmitting “obscene material” to minors online. As Cox wrote in a 2020 reflection, he and Wyden aimed to “protect[] speech and privacy on the internet from government regulation, and incentiviz[e] blocking and filtering technologies that individuals could use to become their own censors in their own households.”

The idea was not to create an entirely unregulated internet, but rather an internet that could regulate itself. Looking back from the vantage point of 2021, there is a certain optimism to the idea that this new online space could work just fine without government interference—that the machine would run smoothly enough that platforms and even users themselves could work out the kinks as needed. The final version of the bill reflects this optimism, describing the internet as “a forum for a true diversity of political discourse, unique opportunities for cultural development, and myriad avenues for intellectual activity” that has “flourished, to the benefit of all Americans, with a minimum of government regulation.”

Two decades later, the picture looks different. In 1995, the year § 230 was drafted, Pew Research Center found that only 21% of Americans had booted up the World Wide Web; 26 years later, in 2021, 93% of Americans told Pew that they used the internet. As experienced by the average user, the internet has shifted from a thicket of small websites to an environment dominated by a handful of enormous social media platforms. Meanwhile, the public perception of the internet is now characterized less by optimism than by trepidation. Both Democratic and Republican members of Congress are pushing for increased regulation of technology companies, though there is little agreement about what form that regulation should actually take. Outside Washington, D.C., the mood is not much brighter. A 2020 study by Gallup and the Knight Foundation found that 60% of Americans felt major technology companies do more to divide society than unite it, and large majorities reported they were “very concerned” about the spread of misinformation and the security of their privacy online.

Over the course of the last decade, social media and information technology has grown into a sprawling, trillion-dollar industry—and many of the current criticisms of social media focus on this massive increase in power and scale. But those criticisms only came to the forefront of public discussion after the 2016 election.

That year, both news organizations and U.S. intelligence agencies uncovered efforts by the Russian government and government-linked actors to use the internet to aid Donald Trump’s presidential campaign and otherwise cause chaos, by disseminating leaked emails hacked from the Democratic Party and trolling on social media. Other reporters chronicled how entrepreneurs from across the world distributed fake news about the U.S. election in order to profit off ad revenue: In one prominent example, an article crafted to look like a real news story claimed that Pope Francis had endorsed Trump’s candidacy. At the same time, support for the Trump campaign from the far right drew wider attention to extremist online subcultures, as mainstream news outlets and even Hillary Clinton’s presidential campaign struggled to explain why and how Trump appealed to participants in hateful corners of the internet.

There are real reasons to doubt how much of an impact any of this online activity had in deciding the election. The political scientist Thomas Rid, for example, argues persuasively in his book “Active Measures” that Russia’s use of social media to post incendiary content was far more effective in shaping American press coverage about Russian election interference than it was in actually changing American votes. Likewise, academic research in the years since 2016 has proven inconclusive as to whether reading or sharing fake news actually changes beliefs or behavior.

But the revelations of 2016 still had a lasting effect. Someone had turned over a rock, and the country was left staring at the worms. As Democratic Sen. Mark Warner put it in a 2017 hearing with representatives from Facebook, Twitter, and Google, “The social media and innovative tools each of you have developed have changed our world for the better. … But Russia’s actions are further exposing the dark underbelly of the ecosystem you have created.”

Scholars and advocates had long raised concerns over the cost of hateful and damaging behavior online. But after 2016, framed as an issue of national security and democratic integrity, that behavior drew far more attention from government and civil society alike.

Scholars and advocates had long raised concerns over the cost of hateful and damaging behavior online. But after 2016, framed as an issue of national security and democratic integrity, that behavior drew far more attention from government and civil society alike. The frictionless flow of information around the globe, it turned out, carried with it falsehoods and hatred. In a leaked 2016 memo, Facebook Vice President Andrew Bosworth reflected, “The ugly truth is that we believe in connecting people so deeply that anything that allows us to connect more people more often is *de facto* good.” Days before Bosworth posted his thoughts, the streaming video service Facebook Live had inadvertently broadcast footage of a user’s violent death; when BuzzFeed News published the leaked memo two years later, Facebook was struggling to navigate a growing network of controversies around misuse of its platform during the 2016 election.

It was in this environment that attention turned to rethinking the responsibility of platforms and the role of § 230. During the same 2017 hearing in which Warner commented on the “dark underbelly” of social media, Wyden—now a Democratic senator—excoriated representatives from the major technology companies:

Now, you’ve discussed your response to these attacks, but it is self-evident, in relation to the power your platforms now have, in the past election you failed. And this is especially troubling because the same Federal law that allowed your companies to grow and thrive, the § 230 law, gives you absolute legal protection to take action against those who abuse your platforms to damage our democracy. … You’ve got the power and Congress has given you the legal protection to actually act and deal with this.

From the vantage point of 2017, Wyden described § 230 in the language not only of entrepreneurship and freedom, but also of civic responsibility and a duty to protect American democracy. Likewise, Cox would later characterize the statute as an attempt to “strike the right balance between opportunity and responsibility.”

This new emphasis on the civic duty of technology companies reflects how the post-2016 techlash changed what the government and the public alike expect from technology companies. As Columbia Law School professor Tim Wu wrote in 2019, the early ideal of an “open and free” internet of largely unregulated speech “has been replaced with a widespread if not universal emphasis among the major platforms … on creating ‘healthy’ and ‘safe’ speech environments online.”

Some scholars had proposed revisions of § 230 before the techlash, though not to the same fanfare. Privacy scholar Danielle Citron, an early critic of expansive judicial interpretations of the statute, suggested in 2014 that “Congress should amend Section 230’s safe harbor provision to exclude the very worst actors.” (In the years since, Citron and I have written together about § 230 and related issues, including on FOSTA.) But criticisms of the statute only really gained steam after 2016, as attention increasingly turned to the harms of the internet as well as its promise. “The Internet’s development over the past twenty years has benefited immeasurably from the immunities conferred by Section 230,” argued Jonathan Zittrain of Harvard Law School in 2017. “We’ve been lucky to have it. But any honest account must acknowledge the collateral damage it has permitted to be visited upon real people whose reputations, privacy, and dignity have been hurt in ways that defy redress.” The same year, technology journalist April Glaser reflected in Slate that “for all it has given us, Section 230 of the CDA has also protected some of the worst parts of the internet.”

But Glaser also noted that the statute’s future was newly up for debate, describing a new effort to amend § 230: The Stop Enabling Sex Trafficking Act (SESTA), an early draft of the bill that would eventually be signed into law as FOSTA. (For simplicity’s sake, I’ll refer to all versions of the legislation as FOSTA.). Americans “are heading into a new era,” Glaser wrote, “one in which Google, Facebook, and others are accepting that they can’t remain immune to liability forever.”

The bill’s origins lay in a probe by the Senate’s Permanent Subcommittee on Investigations into Backpage.com, a classified advertising website that, by the Senate’s estimate, had become the largest and most profitable outlet for posting sex ads in the United States. Adults selling sex consensually often relied on the site as a forum to safely find clients away from the uncertainty and potential violence of the street. In 2017, though, the Senate Subcommittee on Investigations released a report finding that the website also hosted a significant number of ads related to sex trafficking, sometimes involving minors. Backpage regularly edited ads to remove keywords that would identify them as objectionable or illegal, rather than removing them outright—so ads with words like “teenage,” “rape,” and “little girl,” would still be published, just with those words removed.

The stories of young victims of trafficking exploited on Backpage are appalling. Three plaintiffs in a high-profile lawsuit against the site alleged being raped repeatedly as teenagers. One of the young women described her pimp’s enthusiasm for Backpage: Among the site’s “best parts,” he told her, was a feature that allowed users to easily get around prohibitions on sex ads posted by sellers under 18. Backpage, though, was able to successfully shield itself from this lawsuit and others under § 230.

The Senate report, along with outrage over Backpage’s legal successes, turned congressional attention toward shutting down the site. As Kosseff notes in his book, the new information unearthed by the Senate about Backpage’s involvement in editing and soliciting sex ads would arguably have weakened the website’s ability to use § 230 as a shield in the future. But members of Congress argued that additional exceptions to the statute were necessary to “give victims a pathway to justice,” as Republican Rep. Ann Wagner of Missouri put it. In April 2017, Wagner introduced legislation in the House of Representatives to amend § 230 to create exemptions from the liability shield for platforms facilitating sex trafficking. Four months later, a bipartisan group of senators led by Republican Rob Portman and Democrat Richard Blumenthal did the same on the other side of Capitol Hill.

Lawmakers did not initially frame the legislation as part of the broader techlash. For one thing, both the House and Senate bills were introduced in the months before the dam truly broke open on news of Russian use of social media during the 2016 election. It was in September 2017, shortly after Portman and his colleagues put forward the Senate bill, that Twitter and Facebook first began to share data on Russian trolling on their platforms in 2016. And only in November 2017 would Congress hold its first hearings on election interference with representatives from the platforms themselves.

But in a hearing on FOSTA in September 2017, several senators gave voice to the darkening national mood around technology. Portman described sex trafficking as having “moved from the street corner … to the smartphone”—framing the device as a portal into a dangerous online world. Blumenthal spoke of the “social and moral responsibility” of technology companies. “There is a Dickensian quality to the Internet,” pondered Sen. Ed Markey, a Massachusetts Democrat. “It can simultaneously enable and ennoble or degrade and debase.”

There is nothing new about moral anxiety around sex and the internet—§ 230, after all, became law alongside a proposal to criminalize the transmission of pornography to minors online. But this language of responsibility and the duality of benefits and costs—with a new emphasis on the costs—echoed in the conversations that were beginning to happen around technology more generally.

At first, major technology companies opposed FOSTA. But then, in November 2017, something changed. The Internet Association, a lobbying group representing major technology companies such as Facebook and Google, announced that it would support a revised version of FOSTA after all. Facebook Chief Operating Officer Sheryl Sandberg offered a personal endorsement of the bill.

This about-face happened just days after Congress convened its first hearings with representatives from Facebook, Google, and Twitter on Russian use of their platforms for election interference. The technology companies found a frosty reception on Capitol Hill. Some members of Congress used the hearing to hint at—or threaten—the prospect of regulation. Republican Sen. John Cornyn of Texas demanded to know how the platforms “distinguish between how you treat” terrorist recruiters, foreign trolls, and sex traffickers (“Senator, I’d say none of those activities are acceptable on our platforms,” said Google General Counsel Kent Walker) and asked each of the witnesses to “continue to work with us on the Stop Enabling Sex Traffickers Act of 2017”—the name of a Senate iteration of what would become FOSTA.

The Internet Association’s change of heart on FOSTA was a strategic shift, The New York Times reported. Facebook, one of the group’s members, had decided to reverse course on FOSTA after taking a beating during the November hearings. “Perhaps no issue in Washington has exposed the vulnerability of the tech companies” amid investigations into Russian election interference “as much as the sex trafficking bill,” the Times noted in an earlier story. Blumenthal, a cosponsor of FOSTA, told the paper that, “There is much stronger agreement among me and my colleagues that there needs to be more aggressive enforcement action on tech companies like Google.”

At the same time as big technology companies were grappling with their reduced influence on Capitol Hill, an informal coalition of sex workers and activists were organizing against the bill, worrying that FOSTA would endanger those who had chosen to work in the sex trade by leading websites to shut down resources for finding and vetting clients online and communicating with other sex workers. “Make no mistake: sex trafficking is a real, horrible problem,” wrote Alex Andrews, a representative of the Sex Workers Outreach Project, in August 2017. “This bill is not the way to address it.” After the House passed a version of the bill in February 2018, the coalition began an online lobbying effort under the hashtags #LetUsSurvive and #SurvivorsAgainstSESTA.

But the Internet Association’s endorsement was the beginning of the end for FOSTA’s opponents. After multiple rounds of revision, including some more narrowly scoped iterations that ultimately did not make the cut, the bill was signed into law by President Trump in April 2018. Wyden—who had chastised major platforms for not employing § 230’s legal protections to prevent election interference—opposed the legislation to the end, but doubled down on the argument that technology companies should exercise greater stewardship over online spaces. In a speech on the Senate floor against the bill, Wyden pinned the blame on big platforms for the soon-to-be-diminished state of § 230, arguing that their failure to moderate content effectively accounted for FOSTA’s success: “[D]espite the fact that CDA § 230 undergirds the framework of the internet as we know it today, it is on the verge of collapse. That is largely because the big internet companies have utterly failed to live up to the responsibility they were handed two decades ago.”

Wyden was not alone among lawmakers in drawing an explicit connection between FOSTA and a desire for increased interest in the “larger good” on the part of large platforms. Days after FOSTA became law, Facebook CEO Mark Zuckerberg testified before Congress for the first time—answering questions on Russian interference and the Cambridge Analytica scandal, involving data collected on Facebook’s platform that a researcher had improperly sold to a Trump-linked political consultancy in 2016. At the hearing’s beginning, Republican Sen. John Thune of South Dakota told Zuckerberg that FOSTA’s passage “should be a wake-up call for the tech community. We want to hear more, without delay, about what Facebook and other companies plan to do to take greater responsibility for what happens on their platforms.”

“A Very Bad Law and a Very Dangerous Precedent”

Supporters of FOSTA portrayed the legislation as a way to clean up the internet and clear a legal path for survivors of sex trafficking to seek justice against websites like Backpage. But during debates over the bill’s passage, sex workers and internet-freedom advocates argued that FOSTA risked unnecessarily cramping online speech and endangering those working in the sex trade. Congress would have done well to take these concerns more seriously: Three years after its passage, FOSTA does not look like much of a success. There’s no evidence that it helped prosecutors address sex trafficking, as its proponents argued it would—and it may actually have harmed law enforcement investigations of trafficking. It’s placed sex workers in danger and has limited the scope of free expression online.

Three years after its passage, FOSTA does not look like much of a success.

This fallout is in part a result of FOSTA’s confusing drafting. The bill went through repeated revisions as Congress cobbled together different versions of the legislation, resulting in a hodgepodge of a law with a number of moving pieces—few of which are clearly defined. Broadly speaking, the legislation has four main components.

First, it created a new crime, codified at 18 U.S.C. § 2421A, of “own[ing], manag[ing], or operat[ing]” a platform “with the intent to promote or facilitate … prostitution.” Harsher penalties are available if the defendant “promotes or facilitates the prostitution of 5 or more persons” or “acts in reckless disregard of the fact that such conduct contributed to sex trafficking.” Second, the legislation expanded existing federal sex trafficking law, 18 U.S.C. § 1591, to criminalize “knowingly assisting, supporting, or facilitating” sex trafficking.

Third, it created a new exception in § 230 for federal civil claims under 18 U.S.C. § 1595, which establishes a civil remedy for violations of § 1591. The law also establishes a civil cause of action for some violations of § 2421A—but, strangely, does not include a similar §230 carve-out for such lawsuits. It also excludes from § 230 state criminal prosecutions of conduct that would violate § 1591, as well as for conduct violating § 2421A—meaning that states may bring criminal charges under coextensive state law. (Recall that § 230, as originally drafted, permitted federal criminal prosecutions but shielded companies from state criminal liability.) And finally, FOSTA established that state attorneys general may bring federal civil claims under § 1595 as parens patriae—meaning that states may sue on behalf of state residents harmed by conduct violating § 1591. The upshot is that the legislation created a number of new ways for websites to face civil and criminal liability for content related to sex trafficking in both state and federal court.

Many anti-trafficking organizations cheered the legislation as a step forward in cracking down on trafficking and clearing the way for victims to see justice. But others took a different view. “#FOSTA would undermine almost every single thing I would tell people for how to stay alive,” tweeted Kate D’Adamo, a sex-worker-rights advocate and partner at the organization Reframe Health and Justice, as the bill grew closer to becoming law. After the Senate passed the legislation, the Electronic Frontier Foundation, an internet-freedom advocacy organization, announced, “Today was a dark day for the Internet.”

Shortly after FOSTA’s passage, Backpage vanished from the internet—but not because of the legislation. On April 6, 2018, three days after FOSTA had arrived on the president’s desk but before Trump had actually signed it, the federal government seized Backpage.com, shutting down the website. Three days after that, the Justice Department announced the indictment of five Backpage executives and employees and the site’s two co-founders on charges including money laundering and “facilitating prostitution.” Backpage CEO Carl Ferrer separately pleaded guilty to conspiracy charges—again, unrelated to FOSTA.

Nevertheless, news stories and commentary linked FOSTA to the law enforcement actions against Backpage, sometimes hinting at a connection and sometimes drawing it outright. “The seizure of the malicious sex marketplace Backpage.com marks an important step forward in the fight against human trafficking,” said then-Sen. John McCain (R-Ariz.), one of FOSTA’s early Senate sponsors. “This builds on the historic effort in Congress to reform the law that for too long has protected websites like Backpage from being held liable for enabling the sale of young women and children.”

Opponents of FOSTA argued that the Backpage prosecution showed new legislation had not been necessary in order to close down the site. Other websites, though, also began shuttering their services—and pointing to directly to FOSTA as the reason why. Craigslist closed its personal ads section two days after the Senate passed the bill. Under FOSTA, the company explained, the legal risk was too great for it to maintain that corner of Craigslist, on which users often posted solicitations for sex: “Any tool or service can be misused. We can’t take such risk without jeopardizing all our other services, so we have regretfully taken craigslist personals offline.”

Sex workers organizing online tallied other websites that had begun to close them out, either disappearing entirely or limiting access to sexual content. Cloudflare, a web infrastructure company, pulled its services from the social network Switter—which, ironically, had been created as an online refuge for sex workers whose content had begun to vanish from the internet. Terminating service to Switter “is related to our attempts to understand FOSTA, which is a very bad law and a very dangerous precedent,” said Douglas Kramer, Cloudflare’s general counsel, in an interview with the technology publication Motherboard.

FOSTA re-established the moderator’s dilemma resolved by § 230. Platforms that sought to moderate content might risk legal liability if content related to sex trafficking slipped through—a dynamic that could push websites to censor broadly. Or, alternatively, platforms could give up on moderation altogether in an effort to avoid knowledge of material targeted by FOSTA, and thus avoid criminal liability under the law’s standard for intent. FOSTA’s supporters, though, argued that the legislation would have a limited effect, creating legal risk only for genuine bad actors in the internet ecosystem. The law represented “a very narrow carve-out,” said Portman, who sponsored the bill in the Senate.

The problem is that FOSTA “is so badly drafted, no one can agree on exactly what it means,” wrote Daphne Keller, the director of the Program on Platform Regulation at Stanford’s Cyber Policy Center. Consider the language added by FOSTA to 18 U.S.C. § 1591, criminalizing a third party “knowingly assisting, supporting, or facilitating” sex trafficking. But it’s unclear what this means. Imagine that someone posts an ad for sex with a trafficking victim on Craigslist: Would the website face legal liability only if it were aware of the ad itself? Or could Craigslist be at legal risk even if it weren’t aware of the specific ad, on the basis that the site of course knew that it allowed ads in general to be posted? Congress revised an earlier version of the bill to move away from language seeming to criminalize conduct along the lines of the latter example, suggesting that the final law imposes a higher standard for intent—but analysis by Kendra Albert, Elizabeth Brundige, Lorelei Lee, and others indicates that FOSTA’s text may not succeed in implementing that change. The end product is a provision that could potentially criminalize a vast range of conduct.

Also vague is the new criminal statute created by FOSTA, 18 U.S.C. § 2421A, which criminalizes “own[ing], manag[ing], or operat[ing]” a platform “with the intent to promote or facilitate the prostitution of another person.” The problem here, as Albert, Brundige, Lee, and their co-authors note, is that the statute leaves a number of key terms undefined—including “promote,” “facilitate,” and “prostitution.” In a 2020 ruling allowing a First Amendment challenge to FOSTA to go forward, the U.S. Court of Appeals for the D.C. Circuit wrote that “FOSTA arguably proscribes conduct that facilitates prostitution,” including, for example, sex workers sharing resources and harm-reduction information between one another. Section 2421A might describe Backpage’s efforts to keep up ads linked to sex trafficking. But it could also put at legal risk, as the appeals court suggested, “an individual or organization that operates a sex worker-led forum about topics like PayPal.”

The result, in the initial days and weeks after FOSTA’s passage, was that a wide range of companies decided it was simply better to be safe than sorry. In the words of Eric Goldman, “the internet shrank.” Craigslist might have faced legal exposure if ads linked to sex trafficking surfaced on its Personals page. Cloudflare feared the same given its role in providing its content-delivery-network services to a social media platform for sex workers. And other companies made similar calculations.

“We tried in this circumstance to get a law that would make sense for infrastructure companies,” Cloudflare’s Kramer said in his interview with Motherboard, referring to companies—like Cloudflare—providing the services that social media companies and message-boards such as Craigslist use to stay online. “Congress didn’t do the hard work of understanding how the internet works and how this law should be crafted to pursue its goals without unintended consequences.” Kramer’s apparent frustration stemmed from the company’s perception that Cloudflare was being held to the same vague standard as a company like Craigslist, which offers a very different service: Users interact directly with Craigslist in posting and responding to ads, while Cloudflare occupies a position lower in the “stack,” providing a range of infrastructure services that help keep websites up and running. So, could Cloudflare be held liable under FOSTA for content posted on a client website, even if Cloudflare itself never interacted with the content at all? “We’re trying to figure out how law enforcement is going to apply” FOSTA, Kramer said.

Scholars and internet-freedom advocates had raised concerns during FOSTA’s drafting process that the statute’s wording could lead to overly aggressive moderation along these lines. Some also worried that platforms would respond to the moderator’s dilemma by abdicating all responsibility for moderating content. This did happen to some extent: After law enforcement shut down Backpage, replacement websites like OneBackpage.com and Backpage.ly popped up, both hosted in Poland and bearing the disclaimer, “FOSTA/SESTA – No Operator of this site reviews content or otherwise screens the content of the site.” But mainstream U.S.-based companies tended to take the opposite approach and aim for over- rather than under-filtering.

Sex workers had voiced alarm early over this possibility. In the view of activists lobbying against the legislation, FOSTA conflated sex trafficking—that is, a person being forced into sex work against their will—with people working in the sex trade as a matter of deliberate decision. This is not a new concern: Scholars Jennifer Lynne Musto and danah boyd have described how anti-trafficking efforts focused on technology often “blur the boundaries between sex trafficking and sex work.” 

For this reason, Musto and boyd argue, campaigns against online trafficking “risk perpetuating harms against the people they aim to assist.” Much of the campaigning against FOSTA by sex workers focused on how sites allowing online advertisements, including Backpage, play an important role in protecting those in the sex trade. “When I started advertising on Backpage,” wrote sex worker and trafficking survivor Laura LeMoon of her time consensually selling sex, “I felt so safe compared to having to work out on the streets. These websites are a huge harm reduction tool because we only interact with who we want to interact with and we can do it in the safe and controlled environments of our own homes.”

Also concerning was the potential loss of online spaces where sex workers could communicate with one another. The writer Alana Massey, a former sex worker, argued, “The very websites that these bills enable law enforcement to criminalize are precisely where I found the generous communities and actionable advice I needed to get out of and avoid exploitative sex work situations going forward.” Such advice includes “bad date lists”—crowdsourced resources of dangerous clients for sex workers to avoid. In June 2018, two months after FOSTA became law, BuzzFeed News published the story of a sex worker who had been assaulted in her home by a client after she had been unable to check his identity against bad date lists or use other screening mechanisms.

The dangers added up. In September 2018, The Associated Press reported that police departments in Phoenix, San Antonio, and Sacramento had all tallied a spike in prostitution arrests in 2018—consistent with a surge of sex workers newly working on the street after losing access to online portals. San Antonio experienced a 58% increase in arrests from March to August 2018 when compared to the same period the previous year. Multiple sex workers told the newswire that, forced onto the street by the absence of sites like Craigslist and Backpage, they faced uncertainty and violence: “It’s a different kind of danger,” said Kara Alexander, a Florida woman who was raped by a client she had not been able to screen online.

Meanwhile, The San Francisco Chronicle wrote in October 2018 that “Reported crimes related to pimping and sex trafficking have more than tripled in 2018,” and that “officers have made more arrests than previous years as much of the activity had been hidden online” before FOSTA and Backpage’s closure. The deputy director of a local health clinic explained to the paper:

Without being able to advertise online, a huge number of sex workers were forced to go outside, and many have reported that former pimps came out of the woodwork offering to “manage” their business again … The very bill that was supposed to stop trafficking has quite literally given formerly irrelevant traffickers new life.

There is some academic work supporting the idea that the availability of online platforms for advertising limits violence against sex workers. A 2019 paper by economists Scott Cunningham, Gregory DeAngelo, and John Tripp found that Craigslist’s mid-2000s expansion of its “Erotic Services” ad platform into various U.S. cities corresponded with a significant drop in the female homicide rate in those areas—as much as 10% to 17%. Examining only women killed by intimate partners or acquaintances—which would exclude sex workers murdered by unknown clients—there was no equivalent decline in those cities over the same time period. Neither was there a decrease in the rate of male homicides or manslaughters, suggesting that the fall in overall female homicides wasn’t just the result of an overall drop in violence and crime.

Meanwhile, an increasing volume of post-FOSTA research—by both academics and advocacy organizations—indicates that the loss of online resources has hit sex workers hard. A 2018 online survey of sex workers by COYOTE-RI, a sex-work advocacy group based in Rhode Island, found that 60% of respondents said FOSTA had caused them to take on “less safe clients.” Additionally, while 92% had screened clients for safety pre-FOSTA, only 63% were able to after the bill became law. In 2020, 33% of respondents to a nationwide online survey of sex workers by the advocacy group Hacking//Hustling reported an increase in violence against them post-FOSTA. Sex workers surveyed by both COYOTE-RI and Hacking//Hustling said FOSTA had caused significant financial hardship: 45% of COYOTE-RI respondents said they had become “unable to support [themselves]” after the closure of Backpage, and 72.5% of Hacking//Hustling respondents said they were facing increased economic instability after April 2018.

“FOSTA resulted in significant psychological and financial harm,” to male sex workers, according to a qualitative 2019 academic study of men working in the sex trade in the San Francisco area. A number of male escorting and gay dating websites either shut down or sharply limited access after FOSTA’s passage, the study found, though access to male-focused advertising services based in Europe helped some survey respondents stay afloat. Echoing Switter’s experience with Cloudflare, one sex worker said he had lost the personal website and email he used for advertising and communicating with clients after his web-hosting provider suddenly pulled its service with no explanation.

Thanks to the global nature of the internet, FOSTA’s effects stretched far beyond U.S. borders. In New Zealand, where sex work is decriminalized, women in the sex trade reported that the loss of Backpage and other advertising sites forced them to look for work on the street: “It’s just made everything a lot more exploitative,” said one worker named Irene. A regional coordinator at the Aotearoa New Zealand Sex Workers’ Collective complained to researchers about “[t]hese bloody American politicians thinking they know what’s best for the world when it’s us, who work with these women and men who are most affected by it, even here in New Zealand, who have to pick up the pieces.”

Perhaps the most damning report for FOSTA’s supporters, though, came from the federal government itself. FOSTA called for the GAO to produce a report on the law’s implementation three years after the bill became law. Rep. Sheila Jackson Lee (D-Texas), who introduced the requirement, argued that the document would “determine whether victims are actually receiving the relief, assistance, and justice they require” and “tell the story.” But when the office released its findings in June 2021, the results were not encouraging for the legislation’s supporters. “Advocates for sex workers vindicated in Section 230 debate by new GAO report,” read one headline.

According to the GAO, the Justice Department found little use for the new criminal statute codified by FOSTA, § 2421A. The report identified only two cases since 2018 in which the department charged a defendant under § 2421A. In one of the two cases, U.S. v. Palms, the defendant successfully moved for acquittal on that count. In the second, U.S. v. Martono, Wilhan Martono—the owner of CityXGuide, a onetime Backpage competitor—pleaded guilty in June 2021 to two counts, including an aggravated offense under § 2421A.

Why so few cases? It’s true that § 2421A is “relatively new,” as the GAO noted—the federal government has only had three years to make use of it. But the office also writes that “prosecutors have had success using racketeering and money laundering charges against those who control such online platforms in the past.” In other words, § 2421A has not been particularly useful in prosecuting sex trafficking because federal prosecutors simply didn’t need it; they already had plenty of other criminal statutes available. After all, Backpage employees and executives were charged with money laundering and facilitating prostitution under statutes that had been on the books long before the legislation that would become FOSTA was introduced in Congress.

FOSTA’s reporting requirement also instructed the GAO to study criminal restitution under § 2421A(d), which provides for restitution for violations of § 2421A(b)(2)—an aggravated violation of § 2421A “in reckless disregard of the fact that such conduct contributed to sex trafficking”—as well as civil actions brought under § 2421A(c), which allows victims injured by conduct violating § 2421A(b) to recover damages. Unsurprisingly, given that no defendants had yet been convicted of charges under § 2421A, the office found that the Justice Department had yet to seek criminal restitution. The GAO also found that only one person had sought civil damages under § 2421A(c), but the attempt was unsuccessful; in March 2021, a judge dismissed the case.

The GAO report focused only on § 2421A, a result of the limited scope of what Congress asked the office to review. This leaves a number of questions about FOSTA’s efficacy unanswered. Consider, for example, the expanded definition of “participation in a venture” under § 1591: Did it lead to an increase in charges? The statute is consistently the most commonly charged crime in federal trafficking cases, according to yearly data from 2017—the year before FOSTA became law—through 2020. But reporting from the Human Trafficking Institute (on which the GAO relied) shows an overall decrease in trafficking prosecutions since 2015. And analysis of the institute’s data shows little indication of an increase in the use of § 1591 since FOSTA was passed, though the number of defendants charged (if not the number of defendants convicted) increased from 2019 to 2020. Notably, the Justice Department warned during FOSTA’s drafting that the statute’s new definition of “participation in a venture” under § 1591 could “impact prosecutions by effectively creating additional elements that prosecutors must prove at trial.”

Prosecutorial outcomes in federal human-trafficking cases using S 1591, 2017-20

FOSTA also opened the door to state prosecutions for conduct violating § 1591 and § 2421A, but data on state-level criminal charges is diffuse and difficult to tally. As Albert, Brundige, Lee, and their co-authors note in their study of FOSTA, plenty of barriers other than § 230 immunity have prevented states from prosecuting trafficking cases in the past—including the fact that the federal government has typically handled the bulk of trafficking prosecutions. Likewise, it’s not clear whether any state attorneys general have yet brought parens patriae claims under the law.

So, to the extent that FOSTA’s criminal provisions were intended as a punitive measure against sex traffickers, the law seems to have had little effect. The raw number of prosecutions, though, is not the only measure of a criminal statute’s success: Criminal sanctions can also be designed to deter people from engaging in the prohibited behavior in the first place. To that end, former Rep. Martha Roby (R-Ala.), who co-sponsored the House version of FOSTA, argued during the chamber’s debate on the law that FOSTA would not only “hold bad actor websites accountable for these unspeakable wrongdoings,” but also “deter[] websites and individuals from selling human beings online.”

Perhaps, then, the lack of prosecutions under FOSTA might be a success story for the law’s proponents: It could show that traffickers and websites have been scared off. The GAO report didn’t address this question of deterrence directly, but notably, the report contained no data showing that rates of online trafficking—or sales in consensual sex over the internet—have declined. Rather, the GAO described an online sex trade that hasn’t shrunken since April 2018, but instead fragmented across a number of platforms and apps, some of which moved overseas.

And according to the GAO, that fragmentation has actually made it more difficult for the FBI to track down information in sex trafficking cases. The FBI now needs to gather leads across a range of platforms, some of which are encrypted, and struggles to gather evidence from websites and apps in jurisdictions outside the reach of U.S. law. Federal law enforcement’s “ability to identify and locate sex trafficking victims and perpetrators … significantly decreased following the takedown of backpage.com,” the GAO wrote. Backpage had existing relationships with law enforcement and had been willing to cooperate with investigations, unlike overseas platforms.

The GAO’s findings on this point recalled early reporting on local law enforcement’s struggles after FOSTA’s passage and the closure of Backpage. “We’ve been a little bit blinded lately because they shut Backpage down. …We used to look at Backpage as a trap for human traffickers and pimps,” an Indianapolis police officer told a local news outlet in July 2018. He went on: “Ethically, I get it. But it’s not safer for our kids by being shut down.”

This isn’t to say that FOSTA had no deterrent effect. Craigslist and Cloudflare’s decisions to cut their losses demonstrated how the risk of liability under FOSTA discouraged the platforms from continuing to offer services. But FOSTA’s deterrence was unfocused—it’s hard to argue that, by supporting Switter, Cloudflare itself was “selling human beings online,” as Roby described the behavior that she hoped FOSTA would prevent—and harmed many of the same people whom the legislation was ostensibly designed to support. Just after the GAO released its report in June 2021, the popular radio show “This American Life” featured a conversation with a sex worker and former trafficking victim who told the show that FOSTA had changed her life for the worse. She had supported herself through sex work after escaping from trafficking, but after FOSTA, her income dried up; she lost her house, her car, and was forced to drop out of school.

Of course, platforms choosing to avoid potential legal exposure under FOSTA were wary of civil suits as well as criminal liability. And plaintiffs have gradually begun to make use of the new loophole in § 230’s liability shield to bring civil suits against platforms. The statute’s poor drafting has left courts confused over a number of issues, from what specific conduct prohibited by federal sex trafficking law is included in FOSTA’s § 230 carve-out; to whether FOSTA also permits state civil claims concerning sex trafficking; to what, exactly, constitutes “participation in a venture.” Only district courts have weighed in thus far, meaning that there is not yet any binding precedent from appellate courts. But as long as the many ambiguities of FOSTA’s drafting remain unresolved, platforms will have an incentive to remain cautious. “Though the exact legal applicability of FOSTA is speculative,” argue Albert, Brundige, Lee, and their co-authors, “… even the threat of an expansive reading of these amendments has had a chilling effect on free speech, has created dangerous working conditions for sex-workers, and has made it more difficult for police to find trafficked individuals.”

Though the GAO report was limited in its scope and held out the possibility that law enforcement might find more use for § 2421A(b) in the future, sex workers and advocacy groups largely took the report as further evidence that the law had been a mistake. “There was no benefit and everyone said this would happen at the outset,” D’Adamo, the sex-worker-rights advocate, told the publication Filter. Writing in The New Republic, journalist Melissa Gira Grant suggested that “the report might end up pretty much ignored in Congress, even if it commissioned it as a provision of SESTA/FOSTA, as it invalidates its own arguments in favor of the law.”

And, indeed, most members of Congress—including Jackson Lee, Portman, and other backers of FOSTA—have had little to say about the office’s findings. Wagner, the representative who first spearheaded the legislation, weighed in only to object to a critical news story in the publication Roll Call that described responses to the GAO report by FOSTA’s early critics. Wagner argued, without referencing the GAO’s work at all, that “FOSTA-SESTA has had a profound impact on the illegal, online sex trade.” A month later, she suggested in another op-ed—also without noting the GAO report—that FOSTA should serve as a model for future reform of § 230.

“This is a cautionary tale that we have to be very deliberate, thoughtful, inclusive in how we regulate the internet.” — Rep. Ro Khanna

The exception to this silence on Capitol Hill came from Rep. Ro Khanna, a California Democrat whose district includes a swath of Silicon Valley. Khanna voted against FOSTA in 2018 and, a year later, introduced legislation to study FOSTA’s effects on the health and safety of sex workers. “Is anyone actually safer because of FOSTA-SESTA? No,” Khanna tweeted after the release of the GAO report. “Is it time to repeal this law? Yes.”

Khanna’s legislation gained some high-profile support from a handful of members of Congress—including Sens. Elizabeth Warren (D-Mass.) and Bernie Sanders (I-Vt.), along with Wyden. But so far, it doesn’t seem like the GAO’s study has given much of a lift to Khanna’s bill. In a December 2019 interview with The New York Times, Khanna sounded a warning note. Congress “didn’t hear the perspective of the impact it’s having on sex workers,” he told the Times. “This is a cautionary tale that we have to be very deliberate, thoughtful, inclusive in how we regulate the internet.”

Section 230 Reform After FOSTA

As the first major revision to § 230, FOSTA represented a sea change. The foundational statute, once described as the “Magna Carta of the Internet” or “better than the First Amendment,” was no longer untouchable. FOSTA broke down the door that other proposals to amend § 230 are now jostling to cross through. And yet, in the midst of increasing political focus on § 230, FOSTA has remained strangely absent from the discussion, even though it has a great deal to teach about how to evaluate the current crop of ideas for § 230 reform. Or perhaps FOSTA is absent because of what it has to teach: specifically, that Congress should slow down when it comes to § 230.

The law was, in some sense, ahead of the curve. FOSTA’s drafting and passage was buoyed by a wave of post-2016 anti-tech sentiment. But the newly critical cultural posture toward technology and social media looked different in 2017 and early 2018 than it does just a few years later, now reshaped by a Republican Party that is increasingly dismissive of the reality of Russian election interference and galvanized by exaggerated claims of social media censorship of the American right.

During the Senate Intelligence Committee’s first hearing with tech companies in 2017, Democratic and Republican senators sounded notes that were more or less similar. Republicans were more eager to underline that Russian trolls had posted messages appealing to both the right and left, but members of both parties voiced alarm at news of election interference and frustration with technology companies for not doing more. Only one senator brought up § 230: Ron Wyden.

By the fall of 2020, opposition to § 230 was such an important component of Republican political identity that Trump was regularly railing against the statute during campaign rallies for his re-election. When he signed FOSTA in spring 2018, though, he did not mention § 230 at all. It took another two-and-a-half years for him to latch onto § 230 as a political rallying cry. The then-president sent his first tweet about the statute in May 2020—just days after Twitter appended a fact-check to a previous post by Trump casting doubt around the reliability of mail-in ballots used for the upcoming presidential election. In the end, Trump tweeted about § 230 over three dozen times, usually calling for its repeal, until Twitter removed his account following the Jan. 6, 2021, riot at the Capitol.

Trump did not come up with this attack on § 230 himself. If the seeds of right-wing resentment toward tech platforms were planted in the broader post-2016 techlash, they first began to blossom in 2018, just around the time that FOSTA was making its way to Trump’s desk. In April of that year, Republican Sen. Ted Cruz of Texas informed Facebook CEO Zuckerberg during a hearing, “The predicate for § 230 immunity under the CDA is that you’re a neutral public forum”—meaning that, as far as Cruz was concerned, Facebook would lose its § 230 immunity if it chose to moderate content. This is backward, as Cruz—who had clerked on the Supreme Court and served as the solicitor general of Texas—was presumably aware: § 230 was passed precisely so that platforms could enjoy immunity even if they chose to moderate.

This didn’t stop other Republicans from climbing on board with Cruz’s approach. Josh Hawley, then a Republican senator-elect and Missouri’s attorney general, suggested in November 2018 that Congress should “investigate” whether Twitter “target[ed] political speech”—hinting, wrongly, that an answer in the affirmative would require stripping the platform of its protection from liability. “Why do Republicans make fools of themselves by misreading Section 230 of the Communications Decency Act?” asked Nilay Patel, editor-in-chief of the technology website The Verge.

From there, it was off to the races. During the 116th Congress, which ran from January 2019 to January 2021, both Democratic and Republican lawmakers introduced almost 30 different bills to alter or repeal § 230 in some way; over 15 of those bills were proposed solely by Republicans looking to attack Big Tech. Trump’s sudden interest in the statute in May 2020 coincided with his signing of an “Executive Order on Preventing Online Censorship,” which legal scholars brushed off as more of a political performance than an attempt at actual reform.

By this point, the combination of the coronavirus pandemic and Trump’s increasingly fevered efforts to cast doubt on the integrity of the upcoming election had pushed platforms toward more and more aggressive content moderation—often of Trump supporters riffing on the president’s lies about both the virus and the election. The result was a predictable cycle in which Trump or his supporters posted something objectionable; Twitter and Facebook took action against it; and Trump then became even angrier over this “censorship,” complaining more and more about it and further driving outrage on the right. Data compiled by the Internet Archive’s Television News Archive shows a spike in discussions of § 230 on cable news in the summer and fall of 2020, with by far the most coverage on Fox News.

At the same time that anger with § 230 transformed into a right-wing rallying cry, Democrats were developing their own critiques. Where Republican politicians were often irate that platforms had taken content down or otherwise limited access to it, Democratic politicians voiced frustration that platforms were leaving too much content up—misinformation around the coronavirus, far-right extremism, or lies posted by Trump and his associates. As it had on the right, § 230 because a lightning rod for these criticisms. Biden voiced his call for repeal of § 230 during the 2020 Democratic primary when describing his frustration with Facebook’s failure to, in his view, adequately respond to an effort by the Trump campaign to suggest a nefarious connection between Biden and Ukrainian government. Democratic lawmakers, not to be outdone by their colleagues across the aisle, have introduced a number of proposals for § 230 reform as well.

“Just a few years ago, Section 230 was still relatively unknown to the American public,” Wagner wrote in her August 2021 op-ed promoting FOSTA as a model for § 230 reform. She connected FOSTA to the broader frustrations over the internet that had become more common since her legislation became law: Echoing the complaints of her fellow Republicans, she argued that “numerous American citizens have been de-platformed with little explanation and no recourse.”

On both the right and the left, though, Wagner has little company in drawing a link between FOSTA and more recent calls to re-evaluate §230—or even in mentioning FOSTA at all. An archive of congressional press releases compiled by ProPublica shows over 250 statements by legislators on § 230 since the end of 2018—but only about 50 on FOSTA. (Khanna’s press statements challenging FOSTA pop up the most.) Advocacy groups and academics have pointed to FOSTA as an argument for caution in drafting changes to § 230, but by and large, members of Congress remain silent on the subject.

It’s not that the statute has disappeared from the news, however. On the contrary, press coverage became increasingly critical after FOSTA became law. Where early stories about FOSTA tended to omit opposition by sex workers from analysis of the bill, later reporting gave more attention to the concerns of people in the sex trade—a change that The New Republic’s Grant attributed in part to online activism by sex workers.

Perhaps members of Congress haven’t engaged more with FOSTA because the statute is, simply, embarrassing. It’s poorly and confusingly drafted, and as a result, lawmakers—even if they’re not sympathetic to sex workers—have little to show for their work. Though sex workers are increasingly present on Capitol Hill post-FOSTA, they don’t yet make up a constituency Congress feels obliged to answer to, which makes FOSTA’s flaws easier for legislators to ignore. In an interview with “This American Life,” FOSTA co-sponsor and former Sen. Claire McCaskill (D-Mo.) commented that there is “no evidence to say that this bill didn’t do its job. … It’s irritating to me that there seems to be this narrative that sex workers are putting out there [that] well, it didn’t do any good.” She insisted that she “believe[d] it in [her] bones” that the statute had been effective.

Another possibility is that, precisely because it was ahead of the curve on § 230 reform, FOSTA is a poor fit for the current political dynamics around § 230—an awkward precedent that’s useful for nobody in the push for reform. At the moment, the typical complaint expressed by Republicans is frustration that platforms have taken down content that members of Congress would prefer they leave up. They’re looking for platforms to moderate less—or perhaps not at all. FOSTA, though, was driven by a desire for platforms to moderate more. For Republicans, embracing FOSTA as part of their attack on tech platforms would require acknowledging that removing or limiting § 230’s liability protection would likely push Twitter to be more, rather than less, aggressive in moderating their constituents’ posts.

FOSTA also isn’t a helpful example for Democrats seeking more heavy-handed moderation. Democrats have increasingly pushed for platforms to take aggressive action against racist and extremist content and bad information about COVID-19 vaccines. Perhaps because of the newfound cultural focus on § 230, such proposals often center on the statute. But FOSTA’s uneven legacy cuts against the idea that tweaks to the statute represent a quick or uncomplicated fix to the underlying problems—which is not necessarily a message that many Democrats want to hear.

Nevertheless, FOSTA should be an encouraging precedent for supporters of § 230 reform for one major reason: It shows that amendments to § 230 can gain bipartisan backing. In the House, the bill passed 388-25; in the Senate, the count was 97-2. The comparison is not exact. The political environment has changed, and FOSTA focused on a substantive issue, sex trafficking, that was able to draw support from across the aisle, while debate has now centered around § 230 itself. But it’s worth keeping in mind that agreement on § 230 reform may be closer than it seems.

There are currently so many proposals to amend § 230—and new proposals seem to arrive so quickly—that it would be useless to try to summarize them all. But it’s possible to sketch out a few different categories of reform ideas introduced in Congress. Some legislation seems to limit the scope of § 230, carving out exceptions along the lines of FOSTA. The SAFE TECH Act would limit use of the liability shield for, among other things, ads or paid content, as well as civil rights law. The EARN IT Act, a prominent bipartisan proposal that has recently gained momentum in the Senate, would remove § 230 protection for child sexual abuse material and create additional legal risk for platforms employing end-to-end encryption.

Lawmakers have also begun to focus their attention on limiting § 230 protections for the algorithms used by platforms like Facebook and Twitter to sort and recommend content. The Protecting Americans From Dangerous Algorithms Act, drafted by Democratic Reps. Tom Malinowski (N.J.) and Anna Eshoo (Calif.), opens up platforms to lawsuits over civil rights violations or extremist content promoted by certain algorithms. In October 2021, Democratic leadership on the House Energy and Commerce Committee introduced the Justice Against Malicious Algorithms Act, which would remove § 230 protections for “personalized recommendations” provided by platforms’ algorithms that “materially contribute[] to a physical or severe emotional injury.”

Another category of reforms would seek to impose additional obligations on platforms that must be met in order to receive protection from liability. For example, the PACT Act, introduced by Democratic Sen. Brian Schatz (Hawaii) and his Republican counterpart Thune, conditions § 230 immunity on platforms’ adherence to court orders requiring removal of illegal material. And, of course, there are Republican proposals aimed at countering what lawmakers see as anti-conservative bias—which can take the form of bills to condition immunity on new obligations, limit its scope, or even repeal § 230 entirely, but are distinguished by their concern over supposed censorship of users for right-wing political views.

The public conversation around § 230 has become more nuanced since the earliest days of its abrupt prominence; most policymakers, with a few exceptions on the right, are no longer calling for its outright repeal. Notably, the idea of distinguishing between platforms of different scales and functions has become more common. The Justice Against Malicious Algorithms Act includes carve-outs for small businesses, as well as some “internet infrastructure” companies, like Cloudflare. The PACT Act incorporates similar small business and infrastructure exceptions, and the Protecting Americans from Dangerous Algorithms Act adopts a small business carve-out as well.

This represents progress since 2018. FOSTA created chaos online in part because the statute established the same muddled standard of liability across companies that provided a wide range of different services—from Cloudflare to Craigslist. It likewise failed to distinguish between platforms of vastly different sizes, meaning that a service like Facebook—with billions of users and hundreds of millions of dollars budgeted for content moderation—has the same responsibility for eliminating content that could render the platform liable under FOSTA as does, say, a small neighborhood blog or the hiking website AllTrails, which allows users to upload running and hiking routes.

Big platforms have more content to review, but they also have more resources with which to review it. And, importantly, they also have the money to fight any potential criminal or civil litigation under FOSTA, so the risk of leaving content up is lower than it is for smaller outlets, which will face more pressure to shut their doors or cease allowing user comments in order to avoid ruinous legal bills. Facebook’s endorsement of FOSTA, in other words, should not have been reassuring about the effects the law would have on the internet ecosystem more broadly.

Of course, acknowledging these gradations of platform size and function is one thing; determining exactly how regulation should divide platforms up is a much more complicated proposition. What metrics should be used? What levels of responsibility should differently situated platforms have? But it’s at least somewhat positive that serious legislative proposals on § 230 are increasingly incorporating this nuance.

There are other lessons to learn from FOSTA, though, that Congress would do well to pay attention to. Section 230, it turns out, is a sensitive dial: Small adjustments can have wide-reaching, and unanticipated, effects. That doesn’t mean that lawmakers can’t fiddle with the dial at all. They should, however, consider the possible outcomes carefully—and with as much information as possible in hand—before tinkering with the dashboard. The § 230 expert Jeff Kosseff, who testified before the House on FOSTA, said in a 2021 interview that his endorsement of a narrow exception to the liability shield was “probably one of the things I regret the most.” A better approach to revising the statute, he argued, would be “having something with credibility and fact-based, which doesn’t just hypothetically guess at what the harms are.”

Section 230, it turns out, is a sensitive dial: Small adjustments can have wide-reaching, and unanticipated, effects.

In particular, FOSTA’s legacy suggests that Congress should pay special attention to how any proposed changes to § 230 could affect vulnerable constituencies. Right now, women, people of color, and members of other groups subject to harassment and discrimination disproportionately shoulder the costs of § 230. “Women were disproportionately the plaintiffs in the [§ 230] cases with the most disturbing set of facts,” Kosseff writes in “The Twenty-Six Words that Created the Internet.” Mary Anne Franks and Citron, the privacy scholar, have studied how women are targeted online by victimizers who release intimate photos or video of them across the internet or threaten to do so. As Olivier Sylvain, now a senior advisor to the Federal Trade Commission, argues, this kind of harassment “makes online engagement more difficult” for its targets.

But as FOSTA shows, the frustrating irony is that vulnerable people are also likely to suffer from the sort of overly aggressive moderation that could result from sloppy changes to § 230. Marginalized groups can be booted from platforms when companies begin moderating more stringently: Queer people often find their posts removed or banned when a platform decides to crack down on sexual content. And automated content moderation, which platforms have increasingly put to work in recent years, has been shown to perform less accurately on speech by Black users, incorrectly removing their posts as hate speech.

It’s not hard to imagine a world where, after broad carve-outs to § 230, platforms increase moderation at the cost of decreasing Black, queer, and other minority presences online. FOSTA makes this lesson clear—but it’s less clear whether Congress has been paying attention. In June 2021, Democratic Sens. Amy Klobuchar (Minn.) and Ben Ray Luján (N.M.) introduced a bill that would create a sweeping exception to § 230 protections for algorithmically promoted “health misinformation” under certain circumstances. As the pandemic has shown, though, government and medical authorities can provide information about public health that’s less than accurate—recall the early advisories by health organizations that masks did not protect against coronavirus infection. A poorly defined immunity exception along these lines could end up pushing websites to shut down vast swaths of online discussion rather than struggle to navigate the shifting boundaries of what is and isn’t health misinformation.

The problem is particularly pronounced because members of marginalized groups are less likely to participate in the legislative process. Sex workers likewise struggled to have their voices heard in Congress during FOSTA’s drafting. “They were simply invisible,” Khanna, the California representative, recalled in 2021. In the case of Klobuchar and Luján’s Health Misinformation Act, imagine how a future presidential administration could leverage such a tool to limit online information about abortion, contraception, or health care for transgender people. Similarly, many disabled people—a group often pushed to the edge of political conversation and with a strained relationship to medical authorities—use online communities such as Facebook groups to seek and provide medical advice and confer on which therapies are helpful or abusive. Limiting § 230 protections around an ill-defined category of health misinformation could push risk-averse platforms to shut down these conversations.

The Health Misinformation Act is on the more aggressive end of the proposals currently before Congress. By contrast, Schatz and Thune’s PACT Act is likely the most careful intervention, stripping § 230 immunity only from platforms that fail to remove material deemed illegal by a judge within four days of a receiving a court order. The bulk of the bill focuses not on § 230 reform but on instituting, among other things, a range of transparency requirements not linked to liability protection, which would require platforms to provide users with more information about the scope of their content moderation policies and practices. The PACT Act is not perfect, but the cautiousness of its approach is, after FOSTA, arguably a mark in its favor. As misinformation expert Hany Farid of the University of California, Berkeley told The New York Times of § 230 reform, “I think we want to take as modest of a step as possible. Give it a year or two, see how it unfolds and make adjustments.”

Then again, given that lawmakers have shown little interest in revisiting FOSTA, it may be asking too much to expect an increasingly dysfunctional Congress to return to § 230 reform and make intelligent adjustments after watching any amendments play out. Another approach, proposed by Kosseff, would be for Congress to establish a nonpartisan investigative commission to review the mechanics of how companies govern their platforms in order to get a better grasp on which changes to § 230 might be most useful and have the least potential for unforeseen destruction.

Likewise, Stanford Law School professor Nathaniel Persily has argued that, before Congress pushes forward with reform of technology governance, it should first require major companies to provide researchers with access to data about how their platforms curate content and how users engage with it, among other things. Sens. Chris Coons (D-Del.), Portman, and Klobuchar recently introduced legislation modeled on this proposal. Without access to platform data, Persily wrote, “The U.S. government, like its counterparts around the world, is rushing headstrong and blind toward regulation without a complete understanding of the problems they wish to solve.”

But a headstrong rush, unfortunately, has its own political appeal. The early internet-freedom activists conceptualized online space as a separate realm liberated from the constraints and inequities of the offline world. Two decades later, this vision seems increasingly distant. In America, the bookending events of Donald Trump’s presidency—Russian interference in the 2016 election and the Jan. 6, 2021, attack on the Capitol, a riot planned online and fueled in part by falsehoods spread on social media—made clear that the “real” world and the internet are one and the same. Perhaps because this newly widespread recognition of the internet as a central space for civic life also arrived in a moment of political and cultural crisis over what that life should look like, that recognition can lend itself to abstractions of its own. Often, this thinking frames the online world as a uniquely flawed and dangerous place, where all the harms of offline existence multiply and magnify themselves.

Governments have long tried to scrub the public sphere of sex work, pushing people in the sex trade away from respectable areas of civic life. In that sense, FOSTA was nothing new. The civic space that needed to be cleaned up was not a city street or a park but the internet—the new public square. “It was never Congress’s intention … to make a red-light district out of the Internet,” Wagner commented during a White House ceremony for FOSTA’s signing. Writing in 2021, D’Adamo described FOSTA and other broad proposals to reform § 230 as a form of “digital gentrification,” seeking “displacement and disposal of unwanted people, in order to make a space at least look a little nicer.”

A similar anxiety lurked in the sudden post-2016 interest in reining in social media to protect against Russian election interference. This time, the threat polluting civic space was not sex trafficking but the alien presence of foreigners masquerading as Americans on Facebook. At the inaugural 2017 hearing, Senate Intelligence Committee Chairman Richard Burr (R-N.C.) voiced outrage over “agents of a hostile foreign power reached into the United States, using our own social media platforms, and conducted an information operation intended to divide our society.”

By 2020, though, the flood of lies around Trump’s re-election campaign—and the informational chaos of the coronavirus pandemic—pushed the conversation about internet regulation away from protecting against foreign influence and toward how to grapple with harassment and falsehoods coming from within the United States. Arguments between Democrats and Republicans over what content should be taken down are also arguments about who should be allowed entry into the public square. Democrats and many on the left are looking for a civic space scrubbed of bile and falsehoods, even in the absence of a reliable method of distinguishing what should and shouldn’t be permissible; pro-Trump Republicans want a space where they can speak without having to be inconvenienced by the moral judgment of those who find their beliefs offensive.

The tendency to see the internet as an intangible, abstracted public square encourages further abstraction. Proposals for internet regulation often function as a moral or aesthetic statement about what an idealized vision of what civic life could look like. But adjustments to § 230 are a poor means of making that vision a reality: Legislating based on metaphor results in overlooking the very tangible ways that people’s lives, and livelihoods, could be upended. Any serious effort to amend § 230 or otherwise regulate a changing internet will have to set aside symbolism and focus on the frustrating details.

Thanks to Jack Buyske, Evelyn Douek, Emily Hoge, Jeff Kosseff, Alan Rozenshtein, Jacob Schulz, and Benjamin Wittes for their helpful comments and edits.


The Brookings Institution is a nonprofit organization devoted to independent research and policy solutions. Its mission is to conduct high-quality, independent research and, based on that research, to provide innovative, practical recommendations for policymakers and the public. The conclusions and recommendations of any Brookings publication are solely those of its author(s), and do not reflect the views of the Institution, its management, or its other scholars.

Facebook and Google provide general, unrestricted support to The Brookings Institution. The findings, interpretations, and conclusions in this report are not influenced by any donation. Brookings recognizes that the value it provides is in its absolute commitment to quality, independence, and impact. Activities supported by its donors reflect this commitment.

Authors

  • Footnotes
    1. I draw this statistic from the “Section 230 Bill Tracker,” maintained by Alan Kyle.
    2. Both the language that would become § 230 and Exon’s amendment, the Communications Decency Act, were included in the final legislation. Just a year later, the Supreme Court, ruling in Reno v. ACLU, struck down all provisions of the Communications Decency Act except § 230 on First Amendment grounds.
    3. The Australia-based collective Assembly Four, which created Switter, announced in February 2022 that it was shutting down the service, writing, “The recent anti-sex work and anti-LGBTQIA+ legislative changes not only in Australia, but in the UK, US and other jurisdictions have made it impossible for us to appropriately and ethically maintain compliance for over 430,000+ users on a social media platform.”
    4. The litigation, Woodhull Freedom Foundation v. U.S., is ongoing.
    5. boyd spells her name with lowercase letters.
    6. The relationship between sex trafficking and consensual adult sex work is a tricky subject. Contemporary anti-trafficking campaigns tend to collapse the distinction, often as part of an explicit or implicit effort to treat all sex work as inherently nonconsensual and, therefore, something to be abolished. The sex workers whose activism I describe here overwhelmingly reject this argument and take the view that anti-trafficking work along these lines tends to put them at risk.
    7. While the Hacking//Hustling survey collected data online, the group also partnered with the Massachusetts-based group Whose Corner Is It Anyway (WCIIA) to survey street-based sex workers “who face significant barriers to accessing online or digital technology.” Only 19% of respondents to the in-person WCIIA survey (compared to over 70% of respondents to the online Hacking//Hustling survey) said they had faced increased economic instability after FOSTA. However, a comparison of the surveys indicates that the WCIIA respondents were far more likely to be economically unstable before FOSTA and that fewer of them relied on the internet to begin with: 78% of respondents did not have access to a bank account, and only 5.35% reported having regular access to a personal phone over the previous year. Notably, 40% of WCIIA respondents reported seeing more sex workers on the street since April 2018. Hacking//Hustling concluded that “those who are already being heavily policed on the streets do not feel the same immediate effects of FOSTA-SESTA. … What FOSTA-SESTA did was push workers who had access to harm reduction working tools into less safe work environments, increasing their financial insecurity and exposure to violence.”
    8. FOSTA only calls for a GAO tally of cases prosecuted under § 2421A(b), an aggravated violation for “promot[ing] or facilitat[ing] the prostitution of 5 or more persons” or “act[ing] in reckless disregard of the fact that such conduct contributed to sex trafficking.” Perhaps because so few cases had been charged, however, the report includes both U.S. v. Martono—involving an aggravated offense—and U.S. v. Palms, in which the defendant was charged under § 2421A(a). I was unable to find any record of federal defendants charged under § 2421A other than Palms and Martono. (Thanks to Rebekah Carey for research assistance on this point.)
    9. Martono moved to dismiss the § 2421A count on the grounds that the provision was unconstitutionally vague. The judge denied the motion.
    10. Consider FOSTA’s confusing wording concerning § 1595 lawsuits. The statute exempts from § 230 immunity “any claim in a civil action brought under section 1595 of title 18, if the conduct underlying the claim constitutes a violation of section 1591 of that title.” So, would § 230 preclude a lawsuit under § 1595 for conduct not covered by § 1591? This is particularly important because § 1595 establishes a much lower standard for the defendant’s knowledge of the underlying sex trafficking: Defendants sued under § 1595 can be held liable if they “knew or should have known,” but § 1591 requires actual knowledge. An interpretation of FOSTA that strips away § 230’s protection for the lower standard of knowledge under § 1595 alone could be far more permissive and allow a wider swath of litigation. As of early 2022, several district courts have held that FOSTA’s exemption applies only to conduct violating both § 1595 and § 1591. Other district judges, though, have come to the opposite conclusion. Several plaintiffs have argued that FOSTA also permits state civil claims concerning sex trafficking, even though the statute explicitly does not include such claims in its list of § 230 exemptions. At least one federal district court has rejected this argument. But in February 2021, the Texas Supreme Court took the opposite view, agreeing with arguments made by plaintiffs suing Facebook that “by indicating that … section 230 should not be interpreted to bar federal civil statutory human-trafficking claims, Congress must also have been indicating that analogous state civil statutory human-trafficking claims likewise are not barred.”
    11. The statute was also revised in 1998 to require providers to notify customers of the availability of parental controls.
    12. The order called for the Federal Communications Commission to provide its view of § 230 immunity. Biden revoked this request in his own executive order in May 2021.
    13. Though I’ve divided up the various legislative proposals somewhat differently, I’m drawing here on the categories used by the list of proposed § 230 reforms compiled by Slate’s “Future Tense” channel, the Tech, Law, and Security Program at the Washington College of Law at American University, and the Center on Technology Policy at the University of North Carolina at Chapel Hill.
    14. Specifically, under Klobuchar and Luján’s legislation, § 230 would not provide liability protection for health misinformation promoted by any algorithm other than a reverse-chronological ranking, where “health misinformation” would be defined by the secretary of the Department of Health and Human Services during a public health emergency declared by the government.
    15. For example, there is currently widespread debate among autistic people and family members over the use of applied behavioral analysis, a widely used therapy that many autistic people describe as abusive and traumatic. Online arguments on blogs and Facebook groups over use of the therapy have become common.
    16. The bill would also remove § 230 immunity for platforms determined by the Federal Trade Commission to have failed to comply with the legislation’s research requirements, where that failure significantly contributed to the harm alleged by the plaintiff.