Legal Regulation of the Behavioral Impact of AI Technologies on Humans

Abstract

It treats the behavioral influence of AI on humans - algorithmic attention management and the exploitation of cognitive vulnerabilities - as an independent problem that law, built around discrete wrongful acts, is structurally unable to capture; it introduces the category of “cumulative legal invisibility,” by analogy with the evolution of environmental-harm and passive-smoking regulation. A comparative legal analysis of four key jurisdictions (the European Union, China, the United States and Russia) shows that none of the models regulates this influence as an independent subject: the European model comes closest through its ban on manipulative practices (Article 5 of the AI Act), yet covers only the most pronounced forms; Russia’s 2026 law shifted from regulation toward industry support; the American model relies on general law and soft regulation; the Chinese model combines strict content control with state-driven algorithmic value promotion. Building on this analysis, the article proposes five legal solutions: legalizing an objective criterion of behavioral harm, a special regime for emotionally and cognitively interactive systems, recognizing cumulative exposure as an independent object of legal protection, mandatory empirical grounding of regulation, and pragmatic international coordination through technical standards. It concludes that psychological research into AI behavioral effects has applied significance as an empirical basis for operationalizing the proposed legal categories.