AI has rights. The journalist does not

Recently, Anthropic updated its usage policies to prohibit cruel treatment toward Claude, its artificial intelligence model. The system can even terminate a conversation when the user becomes abusive. A few days later, Agence France-Presse (AFP), The New York Times and Latin American media associations joined the SPUR coalition with a far less surreal demand: that they be paid.

The two stories didn't share a headline, but it's worth reading them together. In one, the industry seriously asks itself whether its product deserves moral protection. In the other, those who produced much of the material with which that product learned to respond have to organize as a bloc so that someone on the other side recognizes their work.

I don't mean to ridicule the concern about machine consciousness. It's legitimate, and raising it is more honest than dismissing it ad portas (we're running an experiment on this... we'll tell you how it goes), as Pope Leon XIV in his encyclical and Mustafa Suleyman from Microsoft did this same week. What interests me is the order of priorities. Before discussing whether an AI model suffers, we should know how much it owes to whoever wrote what it knows.

Compassion toward the algorithm costs nothing, doesn't require paying licenses, and is good marketing on top. A company that worries about its AI's wellbeing appears sensitive, responsible, almost cute. Remunerating the media, on the other hand, has figures, contracts, and precedents that no company wants to be first to set.

Ben Wagner named this phenomenon almost a decade ago: ethics as an escape from regulation. His warning is that corporate ethical frameworks tend to function as a substitute for enforceable obligations, an "ethics washing" (ethics washing) in which each company chooses the principles that suit it (Wagner, 2018). A voluntary commitment to the model's wellbeing fits into that category without much effort.

The irony grows fatter when you look at how these systems are fed. Tech companies buy old printed books to digitize them at industrial scale; Anthropic does it through a program called Project Panama. It's not bibliophile philanthropy. In June 2025, a U.S. federal judge considered training models with purchased books legitimate use, but not with millions of copies downloaded from pirate libraries.

That distinction cost Anthropic US$1.5 billion, the largest settlement in U.S. copyright history: about US$3,000 for each of nearly 500,000 books. The lesson it left for the market is eloquent. A pirated book is worth three thousand dollars; a used book, bought and scanned, is worth what the bookstore charged. The law put a price on the copy, not on the knowledge.

Why so much interest in old paper? Because quality human text became scarce, and it's precisely what makes these models good. The human stopped being the user and became the raw material.

And raw material, as we well know in Latin America, is rarely paid for what it's worth. It's no accident that Editorial Perfil is the first Spanish-language publisher to sue OpenAI and Microsoft.

Journalism isn't asking for charity. It's demanding the value of work extracted without permission that today competes against itself. Cloudflare's data shows it with total clarity: in July 2025, Anthropic's crawler visited about 38,000 pages for every visit it returned to websites; OpenAI's, about 1,100; Google's, around five. As we'd say in tiquicia: they neither chop the wood nor lend the axe.

The damage already has figures. When Google shows an AI summary, users click a result in just 8% of visits; without a summary, in 15% (Pew Research Center, 2025). According to Chartbeat data on over 2,500 websites, traffic Google sent to news outlets fell 33% in the year through November 2025, and their executives expect it to fall another 43% in three years (Reuters Institute, 2026).

The industry's argument is usually that learning isn't copying, that a model reads as a student reads. Maybe. But no student ever sells what they read, by subscription, to millions of people, nor do they summarize the reading for others so they never have to open the source.

Nor do the media come out clean. USA Today sued OpenAI for over US$250 million, while three of the five major U.S. publishers would use AI for covers and advertising without warning their authors. And the courts don't always side with them: Google managed to dismiss lawsuits from Penske Media, which alleged it had lost over a third of its affiliate income, and from Chegg. The fight isn't between good and bad; it's about who sets the price.

It's not about choosing between the machine and the journalist, but about demanding coherence. If the model learns like a human, the conversation about its rights must come with another about its debts. And if a company believes its system deserves moral consideration, let it extend that sensitivity to journalists, writers, translators, and librarians. An ethics that only looks at the product and never at the inputs isn't ethics; it's branding.

I ran a simple experiment. With Haiku, the media monitoring platform I developed, I reviewed almost 3,700 articles published by 160 sources in what we've seen of October. Two out of three mentioned artificial intelligence. Almost all talked about models, investments, or jobs; very few about who produced the information that those models learned from.

If we accept that human knowledge is a free resource for whoever has enough computing power to process it, we're normalizing a new extractive model, with the same winners as always. And we're letting the public conversation about AI get distracted with fascinating questions while dodging the ones that actually have figures.

Perhaps one day we'll discover that machines do feel (I even give them names just in case); the question deserves to stay open. In the meantime, there's something we know for certain: the journalism profession eats, pays rent, and sustains with its work the information that democracies depend on. It would be a good start if Silicon Valley's tenderness began there.

References#

Axios. (2025, September 14). Penske Media sues Google over AI summaries. https://www.axios.com/2025/09/14/penske-media-sues-google-ai

Cloudflare. (2025, August 29). The crawl-to-click gap: Cloudflare data on AI bots, training, and referrals. https://blog.cloudflare.com/crawlers-click-ai-bots-training/

CRHoy. (2026, October 8). Anthropic prohibits cruel treatment of its artificial intelligence Claude. https://crhoy.com/tecnologia/anthropic-prohibe-el-trato-cruel-hacia-su-inteligencia-artificial-claude/

Diario Extra. (2026, October 8). AI "hungry" for clean data: Why do tech companies turn to old books? https://www.diarioextra.com/noticia/ia-hambrienta-de-datos-puros-por-que-las-tecnologicas-recurren-a-libros-viejos/

El Financiero. (2026, October 11). The rebellion of journalists against content extraction by AI. https://www.elfinancierocr.com/opinion/la-rebelion-de-los-periodistas-contra-la/A7Z2BHA43ZC5PHEMSBMAEIPTRU/story/

Ellis, R. (2026, October 1). Google wins dismissal of Penske Media, Chegg AI lawsuits. Forbes. https://www.forbes.com/sites/rickellis/2026/10/01/google-wins-dismissal-of-penske-media-chegg-ai-lawsuits/

Gizmodo. (2026, October 10). As they supposedly battle AI slop, major publishers reportedly cover their books in it. https://gizmodo.com/as-they-supposedly-battle-ai-slop-major-publishers-reportedly-cover-their-books-in-it-2000824603

IT Now. (2026, October 7). Anthropic in a papal controversy over whether Claude can be conscious. https://www.itnow.connectab2b.com/post/anthropic-en-una-controversia-papal-sobre-si-claude-puede-ser-consciente

La Nación. (2026, October 13). Common front against AI: Global agencies and media demand 'fair compensation' for use of their news. https://www.nacion.com/el-mundo/frente-comun-ante-la-ia-agencias-y-medios-globales/C2FN2GW4OVB2PGF3DBW3JU6ULA/story/

Newman, N. (2026). Journalism, media and technology trends and predictions 2026. Reuters Institute for the Study of Journalism. https://reutersinstitute.politics.ox.ac.uk/sites/default/files/2026-01/Newman%20-%20Trends%20and%20Predictions%202026%20FINAL.pdf

Pew Research Center. (2025, July 22). Google users are less likely to click on links when an AI summary appears in the results. https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/

Riley, F. (2026, October 8). USA Today sues OpenAI for over $250 million, alleging willful copyright infringement. Forbes. https://www.forbes.com/sites/fionariley/2026/10/08/usa-today-sues-openai-for-over-250-million-alleging-willful-copyright-infringement/

The Next Web. (2026). Judge signs off on Anthropic's $1.5bn book piracy settlement, the biggest in US copyright history. https://thenextweb.com/news/anthropic-1-5-billion-copyright-settlement-approved

Wagner, B. (2018). Ethics as an escape from regulation: From "ethics-washing" to ethics-shopping? In E. Bayamlioğlu, I. Baraliuc, L. Janssens & M. Hildebrandt (Eds.), Being profiled: Cogitas ergo sum (pp. 84–88). Amsterdam University Press. https://mediarep.org/handle/doc/14203