Beyond Editorial Judgement: Social Media Algorithms and Free Speech

Abstract

Engagement-based algorithms on social media platforms present novel legal and constitutional challenges under Section 230 and the First Amendment. Recent decisions demonstrate the unsettled status of these systems, which automatically select, rank, and amplify content based on predictions about what will capture each individual user’s attention and thus increase advertising revenue. While some courts have treated algorithmic curation as expressive speech for purposes of constitutional analysis, significant questions remain about whether these systems should be granted full First Amendment protections and receive statutory immunity under Section 230.

Algorithmic amplification has a greater capacity for harm than traditional editorial judgment. By prioritizing material most likely to sustain a user’s engagement, these systems may repeatedly amplify sensational, emotionally charged, or extreme content without regard to potential consequences. The concern is not merely that a user may encounter one extreme or dangerous post from another user, but that an engagement-based algorithm may continuously push dangerous content onto individual users, over and over again. This repeated and personalized amplification can contribute to individual harms, such as exposure to content encouraging self-harm or other dangerous behaviors, while also producing broader societal effects by reinforcing echo chambers and increasing the visibility of divisive content.

Engagement-based algorithms differ fundamentally from traditional editorial judgment protected by the First Amendment. Editorial judgment has historically been protected when human actors select and organize content to convey a coherent message or viewpoint. In contrast, engagement-based algorithms lack expressive intent, do not articulate an understandable message, and operate continuously and automatically in response to behavioral data rather than deliberate human choice. These outputs are thus shaped substantially by user behavior and optimization objectives rather than the exercise of traditional editorial discretion.

Algorithmic amplification can be viewed as primarily commercial in nature. Platforms design recommendation systems to maximize engagement and, in turn, advertising revenue, not to convey meaning. Extending traditional editorial protections to these systems risks collapsing the distinction between speech and the technological design of social media platforms. The automated, continuous, individualized, and secret nature of these systems further distinguishes them from protected editorial judgment. Accountability, deliberation, and intentionality, the hallmarks of expressive decision-making, are notably absent from engagement-based algorithms. Treating these systems as protected expression blurs the line between conduct and speech and risks granting immunity for activities designed primarily to maximize profits rather than communicate ideas.

A narrow regulatory response can address the harms associated with algorithmic amplification without undermining free expression or the core protections of Section 230. Liability should attach only when platforms materially amplify harmful content through an engagement-based algorithm. Platforms should disclose when feeds are algorithmically curated, offer users the ability to opt out of personalized amplification, and be subject to third-party audits to confirm their compliance. These targeted measures would balance protection of speech with accountability, transparency, and user autonomy.

Suggested Citation

Riley Houldsworth, Beyond Editorial Judgement: Social Media Algorithms and Free Speech, Mich. St. L. Rev. Forum (Sept. 12, 2026).