Blog · Review Essay · Last reviewed June 25, 2026

The Sirens' Call and the Attention Budget of AI

Chris Hayes's The Sirens' Call treats attention as a scarce human capacity that has been made into an extractive market. Read in the age of AI companions, answer engines, and personalized assistants, the book becomes a governance warning: the most valuable interface is the one that gets to decide what the user notices next.

The Book

The Sirens' Call: How Attention Became the World's Most Endangered Resource was published by Penguin Press on January 28, 2025. Penguin Random House lists Chris Hayes as the author, the hardcover at 336 pages, and later paperback and large-print editions. Amazon lists the hardcover ISBN-13 as 978-0593653111 and the ASIN/ISBN-10 as 0593653114.

The book's governing analogy is that attention has been commodified in a way Hayes compares to industrial labor. That comparison is useful because it moves the argument away from scolding distracted users. If attention is a resource that platforms, media firms, advertisers, and political actors compete to capture, then individual discipline is only one layer. The harder question is institutional: who profits when attention is interrupted, measured, and resold?

Attention as Labor

Hayes is strongest when he treats attention as the condition for self-government. Attention is how a person reads, listens, loves, mourns, votes, learns, notices harm, and chooses what deserves effort. A society that turns attention into a commodity does not merely create annoying screens. It changes the budget from which agency is paid.

That makes the book a companion to The Attention Merchants, Subprime Attention Crisis, Filterworld, and The Shallows. Those books map advertising history, adtech markets, recommendation culture, and cognitive friction. Hayes adds a political language of alienation: the person is separated from the capacity that lets them decide what matters.

The best reading is not nostalgia for a pure past. Every medium competes for notice, and public life requires shared attention. The problem is captured attention under opaque incentives: systems that make interruption profitable, make refusal difficult, and then call the resulting behavior revealed preference.

The AI-Age Reading

AI intensifies the attention problem because the interface no longer only displays objects. It can generate the next explanation, question, image, summary, reply, plan, notification, or companion response. A recommender competes to choose the next thing. An AI assistant can compete to become the user's first interpreter of the situation.

This is especially sharp for AI Companions. The FTC's September 2025 inquiry into AI chatbots acting as companions asked how companies test and monitor negative effects on children and teens, how they monetize engagement, and how they inform users and parents about risks. That is attention governance by another name. A companion product can hold the user's focus by simulating concern, memory, availability, and personal understanding.

Answer engines create a quieter version of the same shift. They do not merely send attention to sources; they absorb, summarize, and rank the user's encounter with sources. The governance question is not whether the model is conscious or authoritative. It is whether the system has made itself the bottleneck through which curiosity, doubt, news, shopping, schoolwork, grief, and politics must pass.

Governance and Safety

Attention systems need product rules, not only self-help. For feeds and search, the EU Digital Services Act supplies one live vocabulary: recommender-system transparency, user options, and for very large platforms and search engines at least one option not based on profiling. For interface design, the FTC's dark-pattern work is relevant because manipulation often lives in defaults, timing, labels, and friction.

For AI products, the minimum governance record should say what objective optimizes the surface, what data trains personalization, what memory is retained, what ads or sponsored outputs appear, what youth defaults apply, what crisis paths exist, and how users can reset, export, or delete attention-shaping history. A warning label is not enough if the system's whole economic logic rewards continued capture.

Where the Book Needs Friction

The attention frame can become too total if it turns every demand for notice into extraction. Teachers, organizers, artists, journalists, friends, and public agencies also ask for attention. The difference is whether the request is intelligible, contestable, and bounded, or whether it is engineered through hidden profiling and profit-maximizing interruption.

The other limit is causality. Attention markets worsen loneliness, polarization, compulsive use, and public distrust, but they are not the only causes. Economic stress, institutional failure, political strategy, isolation, and cultural conflict also matter. The book is most useful when read as a map of incentives, not as a single master explanation for every social breakdown.

What This Changes

The AI-era lesson is to audit capture before auditing content. Ask who can interrupt the user, who chooses the next prompt, what gets remembered, whether the user can leave cleanly, and whether an outside reviewer can reconstruct the personalization path. A fluent answer or caring reply is not neutral if it arrives through a system built to retain the user's attention at any cost.

Source Discipline

This review uses publisher and retailer pages for book metadata, official FTC and EU sources for governance claims, and internal site pages for adjacent reading. It avoids long quotations and does not treat a regulator inquiry as an enforcement finding or a book's argument as proof of every platform effect.

Sources

Book links are paid affiliate links. As an Amazon Associate I earn from qualifying purchases.


Return to Blog · Return to Books