“Attention Is All You Need”: The Right to Attention in the Context of AI and LLMs
The merging of mind and machine has already begun. A person with paralysis can move a cursor just by thinking about it. A hearing aid can translate electric pulses into sound. Large language models extend the boundaries of cognition itself, mapping probability distributions over language to predict and generate thought-like patterns [1]. The boundary between attention and computation becomes less visible with every step toward optimization, and in this process, attention is the first component to disappear [2]. The mind is becoming a shared workspace between the human and the machine. Each autoplay, notification, and recommendation functions as a feedback node in a system optimizing for engagement rather than intent. At scale, these systems manage human focus as efficiently as they manage data.
In the context of these rapid technological advancements, it becomes apparent that American law was built to govern information and conduct, not attention. American law can punish lies, regulate data, and limit surveillance, but it has no framework for judging how digital systems shape human perception. The law presumes that users are able to exercise choice, but in reality, structures of modern persuasion undermine any true agency [3]. A right to attention would thus recognize individuals’ power to control how their focus is directed and extracted, and grant individuals’ attentions similar legal protections to their privacy or property. This article argues that courts and legislatures should extend existing legal doctrines to recognize a right to attention and set limits on corporate design practices that override user choice. Courts already distinguish between influence that respects choice and influence that removes it (e.g. time, place, and manner tests, which limit speech when exposure becomes effectively unavoidable). However, these doctrines are not calibrated to the current scale and sophistication of systems designed to capture and redirect attention [4]. A stronger framework is needed to address the contemporary prevalence of mechanisms that co-opt attention before choice can meaningfully occur.
Conceptually, a right to attention has precedent in the common law of nuisance, which protects individuals from intrusions that disrupt their ability to use and enjoy their environment. Courts have debated whether the glow of an electric sign could constitute a nuisance [5], or whether persistent noise and vibration from industry could violate the right to quiet enjoyment [6]. Each time, judges recognized that nonphysical disruptions could still invade the mind’s tranquility [7].
Autoplay, push notifications, and algorithmic feeds occupy mental space the way noise and light once occupied physical space. The difference is that the intrusion occurs at the point where attention is first allocated rather than after it has settled. Even so, the underlying premise holds: People should be protected from environmental intrusions, including those in increasingly invasive digital environments.
In addition to the common law of nuisance, privacy doctrine provides a theoretical basis for a right to attention. In Olmstead v. United States (1928), Justice Brandeis warned that the progress of science would one day give government — and by implication, private actors — means of espionage without physical trespass [8]. His dissent became the foundation for later privacy law, including the tort of intrusion upon seclusion, which rests on the belief that unwanted access to the mind is itself an injury. This logic survives in more recent decisions. The Court in Carpenter v. United States (2018) held that continuous cell-site tracking required a warrant because it revealed “the privacies of life” [9]. Attention, too, would arguably fall under the scope of these “privacies of life.” When design systems track, predict, and exploit how the mind shifts focus, they perform a subtler kind of surveillance — a kind that does not just monitor what users do, but what they drift towards.
First Amendment doctrine, often seen as the barrier to regulating persuasion, in fact contains the seeds of a mechanism to limit machine-based control, as the Court has long recognized that even free speech can be subject to time, place, and manner restrictions when exposure becomes coercive. For example, in *Kovacs v. Cooper *(1949), the Court upheld an ordinance against blaring loudspeakers, stating that one person’s right to speak must give way to another’s right to be left alone. The decision reflects the principle that speech loses its protected character when it overrides one’s ability to decline its message [10]. Similarly, in Lehman v. City of Shaker Heights (1974), the Court allowed a city to prohibit political advertising on public transit because passengers constituted a captive audience unable to avoid the message [11]. Taken together, these cases recognize that liberty involves not only the right to speak but the right to preserve the mental conditions that make listening voluntary. The state’s power to protect unwilling listeners from inescapable noise should therefore extend naturally to protecting users from digital architectures that inhibit the possibility of turning away.
The Supreme Court’s latest decisions around free speech hint at a dawning awareness as well. During oral arguments in NetChoice v. Paxton and Moody v. NetChoice (2024), several justices pressed counsel on the expressive nature of algorithmic ranking [12]. Justice Kagan questioned whether a feed’s ordering logic should really count as speech if it operates autonomously to maximize engagement. Justice Kavanaugh wondered whether the government could regulate such design without dictating content. Though the Court ultimately stopped far short of resolving this issue, simply discussing and framing “feed architecture” as expression revealed the level of power that now sits in the design of attention mechanisms. What it did not confront is the individual’s ability to remain a free agent in relation to their attention within those systems. Though that question likely fell outside the case’s bounds, it should be addressed in broader development of speech doctrine as part of a right to attention.
A right to attention would codify the boundary between persuasion and manipulation and prohibit the intentional, large-scale design of systems that override user choice. Its logic would be familiar and akin to the reasonableness of nuisance laws and time, place, and manner tests. Critics will warn that such regulation threatens innovation or chills expression. But the law has always intervened when private power outpaces the individual’s capacity to defend their peace. The Court’s modern responses to industrial pollution, deceptive advertising, and workplace danger all followed the same pattern: They began with deference and ended in regulation once the damage was obvious. We have arrived at a similar inflection point; it is time for another course correction.
Freedom begins with the quiet rule of the mind. In a time when attention is being parsed into tokens of prediction, exchanged as a lifeless currency in someone else’s world model, a right to attention would protect the space where thought forms, safeguarding the organic judgment that has always forged society.
Footnotes
[1] Markus Huff and Elanur Ulakci, “Judgments of Learning Distinguish Humans from Large Language Models in Predicting Memory,” Nature News (2025), https://www.nature.com/articles/s41598-025-22290-x. [2] Pablo González de la Torre, Marta Pérez-Verdugo, and Xabier Barandiaran,“Attention Is All They Need: Cognitive Science and the (Techno)Political Economy of Attention in Humans and Machines,” AI & Society (2025), https://doi.org/10.1007/s00146-025-02400-z. [3] Aziz Huq, “A Right to a Human Decision,” Virginia Law Review 106, no. 3 (2022), https://virginialawreview.org/articles/right-human-decision/. [4] Kevin O’Neill, “Time, Place and Manner Restrictions,” Free Speech Center, July 9, 2024, https://firstamendment.mtsu.edu/article/time-place-and-manner-restrictions/. [5] “Shelburne, Inc. v. Crossan Corp,” DarkSky, University of Arizona, accessed October 26, 2025, https://findpolicy.darksky.org/case/shelburne-inc-v-crossan-corp. [6] “Sturges v. Bridgman,” Casebriefs, accessed October 26, 2025, https://www.casebriefs.com/blog/law/torts/torts-keyed-to-goldberg/infliction-of-emotional-distress/sturges-v-bridgman/. [7] “Nuisance,” Legal Information Institute, https://www.law.cornell.edu/wex/nuisance. [8] Olmstead v. United States, 277 US 438 (1928). [9] Carpenter v. United States, 585 US ___ (2018). [10] Kovacs v. Cooper, 336 US 77 (1949). [11] Lehman v. City of Shaker Heights, 418 US 298 (1974). [12] “NetChoice, LLC V. Paxton,” Legal Information Institute, https://www.law.cornell.edu/supct/cert/22-555; Moody v. NetChoice LLC., 603 US ___ (2024).