Governing Adaptive News Curation: Sequential Optimization, Cumulative Exposure Allocation, and Societal Accountability

Digital gatekeeping is increasingly performed by adaptive systems that rank, sequence, package, moderate, and sometimes generate news across repeated interactions. This conceptual analysis asks two questions: how should researchers analyze these systems, and who should be responsible for their effects over time? It brings together gatekeeping theory, research on recommender systems and performative prediction, and scholarship on algorithmic accountability, and it makes three contributions. First, it describes curation through six decision functions (source eligibility, agenda and candidate formation, content generation, packaging, exposure allocation, and moderation) connected by a feedback and optimization layer. Agentic curation is treated as a high-autonomy form of adaptive curation, defined by planning ability, authority to act, reach across functions, and delay before human review, rather than as a separate technology class. Second, it defines cumulative exposure allocation as the normalized distribution of weighted exposure across sources, topics, and population groups over time, and it proposes measures of concentration, breadth, repetition, persistence, and disparity. Third, it links each curation function to the actors who control it, the bodies that oversee it, the evidence they need, the standards they apply, and the remedies they can impose. European Union and United States law illustrate the framework. The analysis shifts evaluation away from isolated outputs and short-term engagement toward patterns of visibility produced by a policy over time, and it proposes testable claims about traceability, concentration, disparity, and constrained multi-objective ranking.