Can an Algorithm Commit a Tort? The Circuit Split Over Section 230

Key Takeaways

• A growing circuit split is reshaping the scope of Section 230 immunity. While the Fourth and Ninth Circuits continue to view algorithmic recommendations as protected publisher conduct, the Third Circuit has taken a narrower approach, holding that certain algorithm-driven content recommendations may constitute a platform’s own conduct and therefore fall outside Section 230’s protections.
• Algorithm design and content recommendation practices are becoming a new litigation battleground. Decisions such as Anderson v. TikTok and subsequent cases applying its reasoning suggest that plaintiffs may increasingly challenge social media platforms based on how algorithms amplify, prioritize or distribute harmful content, rather than solely on the underlying third-party speech.
• Forum selection may become outcome-determinative until the Supreme Court provides clarity. With federal courts reaching different conclusions on whether algorithmic recommendations are protected by Section 230, defendants face increased uncertainty and plaintiffs may have greater opportunities to advance defamation and other tort claims in jurisdictions adopting a narrower view of platform immunity.

Section 230 of the Communications Decency Act—often called “the 26 words that created the internet”[1]—has long shielded social media platforms from liability for content posted by their users.

“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.”

In other words, companies that provide internet services like email or a social media platform cannot be sued based on someone else’s expressive activity or content (i.e., third-party speech).[2] Without Section 230, social media platforms likely could not exist in their current form—they would face relentless liability from merely facilitating the acts of their users. Yet, federal courts have interpreted these words so broadly as to protect internet platforms even when used to fund terrorism, disseminate medical misinformation or facilitate human trafficking.[3]

In the thirty years since Congress enacted Section 230, the internet has also changed. The increased use of algorithms to collect, curate and recommend content on social media platforms has some lawmakers eager to amend or even sunset the law entirely.[4] Additionally, federal courts are looking back at their Section 230 jurisprudence and issuing narrower interpretations of the famous 26 words.

Recently, three Circuits have weighed in on whether the use of algorithms by social media platforms to curate and recommend content is protected by Section 230.

The Fourth Circuit Upholds the Status Quo: In MP v. Meta Platforms,[5] the Fourth Circuit upheld the status quo interpretation of Section 230 and confirmed that Meta could not be held liable for the actions of a mass shooter even when the Complaint alleged the shooter was radicalized specifically by content recommended to him by Facebook.[6] The court noted that even though acts such as recommending a group to a user are editorial in nature, “because Facebook has chosen to automate much of its editorial decision-making,” the claims at hand were barred by the broad immunity conferred by Section 230.[7]

The court conceded that “Facebook allows racist, harmful content to appear on its platform and directs that content to likely receptive users to maximize Facebook’s profits” yet decided its hands were tied by its own precedent.[8] The court further claimed it was “not free to disregard Section 230 or to limit its application based on [its] own assessment of the merits of its expansive reach” and this was instead “a question for Congress, not for judges.”[9]

The Third Circuit Breaks Rank: In its February 2024 decision in Anderson v. TikTok, the Third Circuit shook the status quo in a big way. In a short decision, the Third Circuit held that TikTok could be held liable for the acts of its algorithm in promoting lethal content to children.[10] The case involves a ten-year-old who died after attempting the “Blackout Challenge”—a viral phenomenon in which users would “record themselves engaging in acts of self-asphyxiation” and dare viewers to do the same.[11]

The court based its decision on a First Amendment Case, Moody v. NetChoice, in which the Supreme Court held that a platform’s algorithm is the platform’s own “expressive product” for First Amendment purposes and reasoned that for the purpose of Section 230, a platform’s algorithmic compilation of harmful videos could not be third-party speech and was therefore outside the protection of Section 230.[12] The Third Circuit explained that social media platforms should not have the best of both worlds: “they are fully responsible for their websites when it results in constitutional protections, but the moment that responsibility could lead to liability, they can disclaim any obligations and enjoy greater protections from suit than nearly any other industry.”[13] This could be interpreted as a nudge to the Supreme Court that it is time to weigh in on Section 230 and perhaps take a narrower approach than the one long-established in the Circuit Courts.

The Ninth Circuit Holds Back: The Ninth Circuit’s 2026 decision in Doe 1 v. Meta declined to adopt the reasoning in Anderson.[14] This decision demonstrates that the Third Circuit’s reasoning may not have legs in other parts of the country.

The Plaintiffs in this case attempted to hold Meta liable for Facebook’s role in the violent persecution of the Rohingya Muslim community in Myanmar following Facebook’s botched rollout in Myanmar in 2011.[15] The court found that even though the alleged defects still related to Facebook’s “core design as a publishing platform,” Plaintiffs’ “theories of harm depend[ed] on the content of third party posts,” and thus Facebook is shielded from liability by Section 230.[16] The court also expressly rejected the idea that Moody v. NetChoice had any effect on its prior Section 230 jurisprudence and that “algorithmic content delivery” and “recommending content” is just one form of publisher conduct rather than first party expression by the platform itself.[17]

While the majority did not follow Anderson, two concurrences insisted Anderson was decided correctly. One pointed out that together, Moody and Anderson “demonstrate that there is now both an intercircuit conflict and strong tension with a recent Supreme Court case” and urged the Ninth Circuit “to reconsider en banc our precedent extending section 230 immunity to recommendation of content and connections to users.”[18]

A Circuit Split Emerges

Whether the call to action within the Ninth Circuit will be heard or fall on deaf ears remains to be seen, but this “intercircuit conflict” could help nudge the Supreme Court to finally weigh in on the Section 230 issue. In the meantime, Anderson is good law within the Third Circuit and could have major implications for lawsuits against social media platforms which now have additional exposure to liability for the “acts” of the algorithms they employ.

The U.S. District Court for the District of Delaware has already applied the Anderson framework. In Huckabee v. Meta, U.S. Ambassador to Israel and former Arkansas Governor Mike Huckabee filed a complaint against Meta.[19]

The court, following Anderson, held that Meta’s use of “machine-learning algorithms to predict what content will keep and attract users” and its determination of “which posts appear at the top of users’ newsfeed to maximize exposure to certain posts, including advertisements,” were Meta’s own decisions and therefore its own first-party speech as an “information content provider.”[20] Thus, the court found, Meta’s role here was analogous to TikTok’s in Anderson and Meta was not immune from suit under Section 230.[21]

Impact for Platforms and Defamation Plaintiffs

Ultimately, this circuit split presents a ripe question for the Supreme Court: is algorithmic content recommendation and compilation just an extension of traditional publishing protected under Section 230? Or is a recommendation or compilation the platform’s own first-party expression subject to liability?

While Congress has dragged its feet to address the question, courts are beginning to show an openness to revisiting the now 30-year-old text of the statute with the modern activities of “interactive computer services” and the much larger influence they now hold over our lives. State legislatures are beginning to make their own attempts to move the issue forward as well. For now, forum selection will be critical for plaintiffs while platforms must prepare for additional defamation suits to survive the motion to dismiss stage.

Benesch’s Crisis Management & Defamation team continues to actively monitor court decisions and pending legislation regarding Section 230 and its implication for defamation litigation across the country. Benesch client alerts and legal publications are available for you to receive by signing up HERE.

Further Reading

• Recent state court rulings on “addictive or harmful design features” could lead to more cases that circumvent Section 230 immunity.
• The Third Circuit’s reasoning in Anderson may also eliminate Section 230 as additional barrier for cases in which AI tools hallucinate defamatory material.

[1] Kosseff, Jeff. The Twenty-Six Words That Created the Internet. Cornell University Press, 2019. http://www.jstor.org/stable/10.7591/j.ctvr7fcrd

[2] Anderson v. TikTok, Inc., 116 F.4th 180, 183 (3d Cir. 2024).

[3] Gonzalez v. Google, Inc., 282 F. Supp. 3d 1150 (N.D. Cal. 2017) (holding Google had Section 230 immunity and could not face liability for hosting video content published by the terror organization ISIS); Children’s Health Def. v. Meta Platforms, Inc., 112 F.4th 742, 762 (9th Cir. 2024) (holding that Meta was not a state actor subject to First Amendment Constraints and noting that “the ‘legislative grace’ providers enjoy under Section 230 merely affords them the ability to choose whether to suppress certain third-party speech without risking costly litigation”); M.H. On behalf of C.H. v. Omegle.com LLC, 122 F.4th 1266 (11th Cir. 2024) (holding that Section 230 shielded an online video chat website from claims of child sex trafficking).

[4] See S.3546 – Sunset Section 230 Act, https://www.congress.gov/bill/119th-congress/senate-bill/3546/text

[5] 127 F.4th 516, 520 (4th Cir.), cert. denied sub nom. M. P. By & Through Pin v. Meta Platforms Inc., 146 S. Ct. 287, 223 L. Ed. 2d 10 (2025).

[6] “M.P. contends that Facebook is partially responsible for Dylann Roof’s murder of nine innocent people, including her father.” M.P. by & through Pinckney v. Meta Platforms Inc., 127 F.4th 516, 522 (4th Cir.).

[7] Id. at 526.

[8] Id. at 525.

[9] Id. at 527 (citing Zeran v. America Online, Inc., 129 F.3d 327 (4th Cir. 1997)).

[10] 116 F.4th 180 (3d Cir. 2024).

[11] Id. at 181.

[12] Id. at 184 (citing Moody v. NetChoice, LLC, 603 U.S. 707 (2024)).

[13] Id. at 184 (citing Doe Through Roe v. Snap, Inc., 144 S. Ct. 2493, 2494 (2024)).

[14] 174 F.4th 1159 (9th Cir. 2026).

[15] Id. at 1163.

[16] Doe 1 v. Meta, 174 F.4th at 1167–1168.

[17] Doe 1 v. Meta, 174 F.4th at 1169 (citing Dyroff v. Ultimate Software Grp., Inc., 934 F.3d 1093, 1098 (9th Cir. 2019)).

[18] Id. at 1174.

[19] No. CV 24-773-GBW, 2025 WL 1744357, at *1 (D. Del. June 24, 2025).

[20] Id. at *5.

[21] While Huckabee’s claims broke through the Section 230 barrier, the District court initially held that he failed to properly allege his claims. However, following an appeal to the Third Circuit, some of Huckabee’s defamation claims now survive following Meta’s Motion to Dismiss. Huckabee v. Meta Platforms, Inc., No. CV 24-773-GBW, 2026 WL 2349690, at *4 (D. Del. Aug. 13, 2026). This demonstrates that it may be challenging to adequately plead that a social media platform had knowledge of defamatory content published on its platform or intent to distribute that content, but ultimately a plausible allegation can be made.