Section 230 of the Communications Decency Act of 1996 is twenty-six words long in its operative core: "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." Those words, codified at 47 U.S.C. § 230(c)(1), are the legal foundation on which every modern user-content platform was built. Without them, the modern internet - social media, review sites, search engines, e-commerce marketplaces, message boards - would not exist in anything like its current form.
Section 230 also contains a second, less-discussed provision at § 230(c)(2) that immunizes good-faith moderation. Together, the two provisions resolve a problem that the pre-1996 case law had created: platforms that moderated user content (Stratton Oakmont v. Prodigy, 1995) were treated as publishers and were liable for everything they hosted, while platforms that did not moderate (Cubby v. CompuServe, 1991) were treated as distributors and escaped liability for content they had no knowledge of. The pre-1996 rule punished moderation. Section 230 inverted that incentive.
This guide explains what Section 230 actually does, what it does not do, the leading cases that have shaped its scope, the FOSTA-SESTA carveout of 2018, the 2023 Supreme Court cases that left it intact, and the reform proposals that have been introduced and stalled in every Congress since 2018.
What § 230(c)(1) actually says and does
The full text of § 230(c)(1) is: "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." Three terms carry the weight. "Interactive computer service" is defined broadly at § 230(f)(2) and covers essentially any service that enables computer access by multiple users to a server, including ISPs, hosting providers, social media, search engines, and review sites. "Information content provider" is defined at § 230(f)(3) as any person responsible in whole or in part for the creation or development of the information. "Publisher or speaker" is the common-law term that defines who is liable for defamatory or otherwise tortious content.
The combined effect is that a platform cannot be sued in tort for content created by its users. A defamation suit against Google for a defamatory user review on Maps fails at the motion-to-dismiss stage. A suit against Yelp for a defamatory user review fails. A suit against Facebook for a defamatory user post fails. The platform is not the publisher or speaker; the user who posted the content is. Liability lies with the user, not with the platform.
The immunity is procedural as well as substantive. Courts have consistently held that § 230 questions are decided on the pleadings, often on a Rule 12(b)(6) motion to dismiss, before discovery. Platforms therefore avoid not just liability but the cost of defending the litigation through trial. This procedural feature is as important to the practical operation of § 230 as the substantive immunity itself.
What § 230(c)(2) actually says and does
Section 230(c)(2)(A) provides: "No provider or user of an interactive computer service shall be held liable on account of 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."
This is the moderation immunity. It permits platforms to remove user content they consider objectionable without becoming liable to the user whose content was removed. The phrase "otherwise objectionable" has been read broadly by most courts (Zango v. Kaspersky, 9th Cir. 2009, applying the immunity to anti-spyware blocking) and is the textual basis on which platforms can remove harassment, hate speech, misinformation, and similar content without facing wrongful-removal suits.
Section 230(c)(2)(B) extends the immunity to providers of filtering tools and to users who use those tools to filter their own consumption of content. This provision is the basis on which parental-control software, ad blockers, anti-spam tools, and similar consumer products have been able to operate without facing breach-of-contract or tortious-interference suits from the publishers whose content they block.
What Section 230 does not do
Section 230 does not immunize the original speaker. The user who posts defamatory content remains liable in tort. § 230 protects the platform; the speaker has no protection from § 230 and faces ordinary defamation, harassment, or other tort liability.
Section 230 does not preempt federal criminal law (§ 230(e)(1)), federal intellectual-property law (§ 230(e)(2)), the Electronic Communications Privacy Act (§ 230(e)(4)), or, since FOSTA-SESTA in 2018, federal sex-trafficking law and parallel state laws (§ 230(e)(5)). DMCA copyright takedowns, trademark suits, and federal criminal prosecutions are unaffected by § 230.
Section 230 does not immunize content the platform itself creates. Where a platform authors content (Knowledge Panel summaries written by Google, editorial content on a news site, the platform's own marketing copy), § 230 does not apply because the platform is the information content provider for that content.
Section 230 does not immunize material development of unlawful content. Roommates.com (9th Cir. 2008) held that where the platform's design materially contributes to the alleged unlawfulness (a roommate-matching site that required users to specify discriminatory housing preferences in violation of the Fair Housing Act), § 230 does not apply. The standard is high; ordinary indexing, ranking, and surfacing of third-party content do not count as material development.
The leading cases
Zeran v. America Online (4th Cir. 1997) is the foundational case. The plaintiff was the target of a malicious AOL message-board hoax that listed his phone number as the contact for offensive merchandise. He sued AOL on a distributor-liability theory after notice. The Fourth Circuit held that § 230 forecloses both publisher and distributor liability and that the platform retains immunity even after notice of the unlawful content. Every federal circuit has subsequently followed Zeran.
Fair Housing Council v. Roommates.com (9th Cir. 2008) is the leading case on material development. The court held that § 230 does not apply where the platform's design itself contributes to the unlawfulness, but applied the immunity to the portions of the site that merely hosted user content.
Barnes v. Yahoo! (9th Cir. 2009) carved out the contract / promissory-estoppel exception. Where a platform makes specific representations to a user about content removal and then breaches those representations, contract or promissory-estoppel claims have survived § 230 motions on the contract theory (not on a defamation theory).
Hassell v. Bird (Cal. 2018) held that § 230 prevents enforcement of a defamation judgment against the speaker via removal order against the platform. The platform retains its immunity even after the underlying speaker has been adjudicated to have defamed the plaintiff.
Force v. Facebook (2d Cir. 2019) applied § 230 to algorithmic recommendation of third-party content (terrorism-related claims under the Anti-Terrorism Act), foreshadowing the Supreme Court's eventual non-engagement with the algorithmic-recommendation question in Gonzalez.
Gonzalez v. Google and Twitter v. Taamneh (2023) were the most-watched § 230 cases of the modern era. Taamneh resolved on aiding-and-abetting grounds without reaching § 230, and Gonzalez was remanded in light of Taamneh without a § 230 ruling. The combined effect was to stabilize § 230 in the short term.
Section 230 protects the platform; the speaker has no protection from § 230 and faces ordinary defamation, harassment, or other tort liability. The realistic defendant is the original speaker, not the platform.
FOSTA-SESTA: the only successful carveout
The Allow States and Victims to Fight Online Sex Trafficking Act (FOSTA) and Stop Enabling Sex Traffickers Act (SESTA), enacted together in 2018, added § 230(e)(5), which excludes federal sex-trafficking claims under 18 U.S.C. § 1591 and parallel state laws from § 230 immunity. This is the only successful narrowing of § 230 since the original 1996 enactment.
FOSTA-SESTA's practical effect has been heavily debated. Supporters point to the shutdown of Backpage.com (which had already been seized through criminal forfeiture before FOSTA) and to a small number of civil suits that have proceeded under the new exception. Critics point to the closure of harm-reduction resources for sex workers, the elimination of personals categories on mainstream sites (Craigslist Personals shut down in March 2018 the day FOSTA passed), and the lack of evidence that the law has reduced trafficking. A 2021 GAO report found limited evidence of impact in either direction.
From a doctrinal standpoint, FOSTA-SESTA established that Congress can carve out specific categories from § 230 if it chooses to. Multiple subsequent reform proposals (Justice Against Malicious Algorithms Act, EARN IT Act variants, SAFE TECH Act, the proposed bipartisan repeal of § 230 immunity for algorithmic amplification) have used FOSTA-SESTA as the procedural template, but none has passed.
The reform debate and why it stalls
Section 230 reform is one of the few topics on which both major US political parties have proposed changes - but for opposite reasons. Conservative critics typically argue that platforms moderate too aggressively against viewpoint diversity and that § 230(c)(2)'s good-faith requirement should be tightened. Liberal critics typically argue that platforms moderate too little against harassment, hate speech, and misinformation and that § 230(c)(1) should be narrowed for algorithmically-amplified content.
The structural problem with reform is that the two critiques pull in opposite directions. A reform that limits moderation immunity to address the conservative concern would expose platforms to wrongful-removal suits and incentivize them to moderate less, which directly worsens the liberal concern. A reform that narrows publisher immunity to address the liberal concern would expose platforms to defamation, harassment, and similar suits and incentivize them to moderate more aggressively, which directly worsens the conservative concern. No reform proposal since 2018 has resolved this tension.
Pending federal proposals as of mid-2026 include narrower variants (carveouts for specific harms like deepfake pornography and pediatric self-harm content) that mirror the FOSTA-SESTA template and are more likely to pass than broad reforms. State-law experimentation (Texas HB 20, Florida SB 7072, struck down in part by Moody v. NetChoice in 2024) has tested the constitutionality of state-level platform-content regulation, with mixed results.
What Section 230 means for reputation victims
For an individual or business whose reputation has been damaged by online content, the practical implication of § 230 is that the realistic defendant is the original speaker, not the platform. Defamation suits against the platform fail at the motion-to-dismiss stage in nearly every reported case. Suits against the speaker proceed normally and remain viable.
The realistic remedies that operate alongside the litigation track include DMCA takedowns for content infringing the victim's copyrighted material (not preempted by § 230), GDPR delisting requests for EU residents (operates on a different legal framework that § 230 does not affect), Google's expanded 2022-2024 personal-information-removal policy (a voluntary platform program that requires no legal basis), platform-specific content-policy reports, and SEO suppression to push the content down in search results.
The combined approach is more effective than any single channel. Suing the speaker addresses the source. Platform-specific removal channels address visibility. SEO suppression addresses what cannot be removed. Suing the platform almost never works and almost never produces the practical outcome the victim is seeking.
