Big Tech Litigation Tracker

The §230 Result column is a classification made for this page, not a label any court applied to itself. Each row's full ruling text is preserved verbatim in the expanded detail view, and the short value is derived from that text mechanically — by ordered prefix rules in the conversion script, never typed per row — so the summary cannot drift from the prose it summarises. Seven values are possible: immunity granted, immunity denied, partial denial, not reached, not applicable, contested / undecided, and unverified. Open a row and compare the ruling text against the label if you doubt any single classification; that is what the detail view is for. ↩

These are almost entirely the Anti-Terrorism Act cases — Fields, Crosby, Colon, Retana, Clayborn, and ultimately Gonzalez and Taamneh themselves. They matter to the argument precisely because they are not Section 230 wins for the platforms: the courts never got that far, dismissing on the ATA's proximate-causation and aiding-and-abetting elements first. This is the stronger form of the point. A victim of promoted terrorist content frequently cannot obtain a Section 230 ruling at all, favourable or otherwise, because a different doctrine disposes of the case earlier. Anyone citing "platforms keep winning on 230" should know that in seven of these cases the immunity was never actually tested. ↩

Read this as a statement about a small decided sample, not a law of nature. Eight attorney general rows appear on the table, but only four have produced a dispositive ruling on the defence; the rest settled before one, or are still being briefed. Three of those four denied immunity and the fourth (California v. Meta) allowed the COPPA and consumer-protection claims through in part. A single adverse appellate ruling would change the sentence, and three of the eight AG rows rest on press releases rather than orders — see the flagged list above. The claim is that no attorney general action on this table has yet lost on Section 230, which is a narrower and more defensible thing than "attorneys general cannot lose." ↩

The denominator is the eighteen private rows in category A. They break down as seven immunity granted, seven never reaching Section 230, two partial denials (Doe #1 v. Twitter and In re Facebook, each surviving only on a narrow statutory sliver), one immunity denied (Anderson), and one unverified (Smith v. TikTok, counted in the denominator but not claimed as a loss, since its status has not been confirmed). Filter the table to Private plaintiff plus A — Promoted illegal content to reproduce the count. ↩

Anderson v. TikTok, 116 F.4th 180 (3d Cir. 2024) — the precedential opinion holds that algorithmic curation of a For You Page is the platform's own first-party expressive activity, and so is not "information provided by another" within §230(c)(1). The most important open question on this page is what happens to it next. The decision only reversed a dismissal; it was remanded to the Eastern District of Pennsylvania and has never been tested on the merits, so no plaintiff has yet recovered on the theory. The remand docket is the single most valuable thing to watch, and a Supreme Court grant on the resulting circuit split — against the Second (Force), Fourth (M.P.) and Ninth (Doe v. Grindr) — would be the most consequential development possible for everything on this table. ↩

The judgment itself is held with the site's research materials rather than being quoted from press coverage, and the New Mexico Department of Justice announcement summarises it. One qualification belongs with any use of this figure: the court declined to order hard age verification, holding that "sweeping policy choices of that kind belong to the legislative and executive branches, not the courts." New Mexico is therefore not a total victory even on its own terms — the presiding judge said in substance that the remedy is incomplete without legislation, which is the argument this site makes. ↩

Quoted from contemporaneous reporting of the 14 July 2026 dismissal (9to5Mac and others), not from the order, which has not been pulled. Given how much rhetorical weight a federal judge telling Congress to fix the statute can carry, the order in Doe v. Apple, No. 24-cv-05107-NW (N.D. Cal.), should be obtained from PACER or CourtListener and the wording checked before the sentence is quoted in advocacy or testimony. ↩

The quotation comes from the Minnesota Attorney General's own announcement of the 23 March 2026 ruling, which is a party's characterisation of its own win and not a neutral report. The underlying order of the Hennepin County District Court has not been read. Treat the wording as provisional: it is quoted here because it states the design-versus-content distinction more crisply than any other source found, which is exactly why it should be verified before it is relied on. ↩

Both counts are approximate and drawn from commercial multi-district-litigation trackers rather than the courts. They also move constantly — MDL 3047 was reported as growing by roughly 230 cases in a matter of weeks. The authoritative figure is the Judicial Panel on Multidistrict Litigation's own monthly pending-case statistics, which is where these numbers should be refreshed at each review. ↩

Patterson v. Meta Platforms, Inc. is the sharpest illustration of why every row here carries a date. Justice Feroleto allowed claims against YouTube and Reddit to proceed in March 2024; the Appellate Division, Fourth Department reversed on Section 230 grounds in July

2025. Whether the New York Court of Appeals has granted leave to appeal is unknown to this page, and the Fourth Department's opinion has not been read — both are listed among the unsourced rows above. Anyone extending this table should resolve that first: it is the broadest defendant list of any private suit here, naming Meta, Google, Amazon, Reddit and Discord together.

↩

The search covered Microsoft, LinkedIn and Xbox against content-harm, grooming and exploitation claims. A targeted search is not an exhaustive one, and a negative result of this kind is weaker evidence than a positive one — it establishes that no such suit was prominent enough to surface, not that none exists. It is recorded as a row rather than an omission so that a reader who wonders why the largest software company is missing gets an answer instead of a silence. ↩

For the terrorism cases specifically, the George Washington University Program on Extremism maintains a fuller enumeration of terrorism lawsuits against social media companies than the seven representative Anti-Terrorism Act entries kept here. For new filings and ongoing status, the Tech Justice Law Project's litigation tracker, hosted at Tech Policy Press, is the best continuously-updated source and is where this page should be refreshed from. For doctrinal analysis of individual Section 230 rulings, Eric Goldman's Technology & Marketing Law Blog covers nearly every decision on this table, usually within days, and is cited as the source for several rows — with the caveat that it is openly critical of attempts to narrow the immunity, and should be read as informed advocacy rather than neutral reporting. Federal dockets themselves are free on CourtListener. ↩

litigation.csv, published in this page's own directory, is the human-editable source of truth; litigation.json, which the table actually renders, is generated from it by themes/quark2/tools/csv-to-json.py and should never be hand-edited. Both are committed, so the table can be audited or rebuilt from the repository alone. A working copy of the same research also sits with the primary case documents in the site's research directory, outside version control — it will drift, and the published CSV wins. ↩