In this article
Today we get into an uncomfortable subject, and I'll start by apologising in advance to the Spanish journalists who do good work, because the headline is going to flatten the nuance. The thesis: Spanish media coverage of artificial intelligence is, taken as a whole, well below the level the sector demands. Is it the fault of individual journalists? Almost never. Is it the fault of the newsrooms, the business models, the quick translation and the advertising dependence? Far more. My opinion is biased, because I've been reading tech press for twenty years and the Spanish stuff forces me to read in English too just to not come out lopsided. Form your own: open the technology supplement of any of the five major dailies and count how many pieces are analysis and how many are amplified press release.
For months I've kept a notebook where I jot down, week by week, the AI pieces published in the technology supplements of El País, El Mundo, ABC, La Vanguardia and Expansión. The rough proportion I keep finding is this. Sixty per cent are quick translations of news that came out the day before in The New York Times, the Financial Times, Bloomberg or The Wall Street Journal. Twenty per cent are amplified corporate communications — interview with the CEO of the moment, product launch, investment announcement. Fifteen per cent are news pieces with no analysis: "X has announced," "Y has said that," "Z dropped on the stock market." And five per cent, generously, are pieces with original reporting, sourced from multiple angles, with a critical voice and technical verification. That proportion is the problem, and we're going to take it apart.
Why call it "very bad" and not just "bad"
"Very bad" sounds harsh and needs a defence. The defence is comparative. Take any random week of 2025 and compare the AI coverage of the Financial Times, The New York Times, The Atlantic, Bloomberg or even specialist outlets like The Information, Stratechery, Platformer or 404 Media, and the gap with Spanish general-interest press is an order of magnitude. Not in tone, not in sensibility, not in ideological slant — there each outlet picks its own, and that's no failing. The gap is in the presence of verifiable figures, in technical sources cited by name, in the historical context of the news, and in the clear separation between what is information, what is opinion and what is advertising.
That gap isn't new. We've had the same pattern for at least a decade. But in AI it shows more, because the sector moves fast, the figures are large, the names are foreign and the concepts are technical. When a sector meets those four conditions, mediocre coverage stands out far more than in national politics or sports, where the reader already has prior context.
The structure of the problem, not the people
To make my point clear, it helps first to discard the easy reading. It isn't that Spanish journalists are lazy. The vast majority are competent and work more hours than they get paid for. The problem isn't in the individual. It's in the structure of the section.
The technology desks of the five big Spanish dailies employ, according to the Informe anual de la profesión periodística published by the Asociación de la Prensa de Madrid, between one and three full-time journalists per masthead. That staff simultaneously covers phones, computers, video games, cybersecurity, AI, social media, digital platforms, telecoms, and sometimes startups and entrepreneurship too. A single journalist with that editorial load can't go deep on any of it. Their daily operational task is to close one or two news headlines and keep up the pace of online updates. The analysis piece, the original reporting, the reading of the technical paper, the interview with the MIT researcher who didn't reply on the first try and gets a second attempt later: all of that takes days or weeks that don't exist in the planning.
When the editor asks the journalist to close the piece in four hours because the news is already out, what comes out is not journalism. It's translation and summary. And that, however skilfully done, adds no value over the original source. Whoever wants to understand the news will read the source. Whoever can't read the source will get the trimmed version with no idea what was lost.
The business model that selects
There's a second structural factor, and it's the business model of the technology supplement. Sector advertising is a major revenue source for those sections, and the big advertisers are precisely the companies the section covers. Microsoft, Amazon Web Services, Telefónica, IBM, Salesforce, Oracle, Indra, Google, Meta, Apple, NVIDIA, Samsung. The list is predictable. It isn't explicit corruption, nor a directive sent down to the newsroom. It's soft selection by the ecosystem. What gets covered with emphasis tends to be what the sponsor values; what gets criticised hard tends to be what it doesn't sponsor. There are exceptions, but the statistical pattern holds.
The consequences of the model are three. The first is that CEO interviews are accepted by default when they're offered, with no critical conditions, because the interview is easy content well regarded by the sponsor. The second is that the press release for a corporate announcement comes in as news with barely any rewording, because the journalist hasn't time to find a second source and because the release is already "cooked" for publication. The third is that critical analyses of the sector are less frequent and less visible than the praise. Criticism exists, but it usually sits in shorter pieces, worse positioned on the front page, or reserved for editorials signed by columnists with autonomy. The difference between critically covering Telefónica in a paper that runs Telefónica ad inserts every week and critically covering a Chinese company with no presence in the Spanish market is enormous. And it shows.
Quick translation as a genre
There's a genre of its own that deserves a name. I'll call it quick translation. It works like this. At eleven in the morning, Spanish time, Bloomberg or The New York Times publishes an important story: a new model, mass layoffs at a lab, a ruling, a licensing deal. The Spanish newsrooms spot the story, translate it into Spanish, give it a shorter headline and publish it on their site between two and four in the afternoon. The piece runs 350 to 600 words. It includes no quotes from original sources. It mentions no historical context. It doesn't qualify the figures with footnotes. And, most serious of all, it doesn't explicitly link to the original source so the reader can go deeper.
That genre has a function. It lets the masthead keep up the pace of the news cycle without investing in reporting. But its informational value is very low, because any interested reader has already read the original or will read it as soon as they discover it exists. The translated piece lives off the inertia of the reader who doesn't go looking for the source. And that inertia is legitimate, but the reader ought to know they're consuming a derivative, not a product.
The few who do
There are journalists working against this current. It matters to name them, because they prove it can be done, within the same constraints, differently.
Manuel G. Pascual at El País has spent years covering AI with technical rigour, sourced from multiple angles, with a voice of his own. Marta Peirano — in her book El enemigo conoce el sistema (Debate, 2019) and in her later output — has sustained a systematic critique of the platform economy that the general conversation needed. Jaime García Cantero has brought data analysis into science communication with seriousness. José Mendiola, for years at Engadget and other outlets, kept up technical coverage with judgement. The Maldita.es team, especially Maldita Tecnología, does sector fact-checking that fills the gap the general-interest newsrooms don't. Newtral and, on a smaller scale, some independent newsletters such as Error 500 (Antonio Ortiz) or Café Ambrosía have held an analytical focus.
The list isn't meant to be exhaustive and I'm surely leaving out important names. The list is meant to show that examples of responsible coverage exist, holding out against a structure that tends to flatten them. That their bylines be visible when they are, and that their effort be recognised, is no favour: it's the survival of a genre the Spanish media ecosystem needs.
What happens when coverage fails
It's not just a professional problem. It's a civic one. When most of the public informs itself about AI through a layer of mediocre coverage, the consequences show up in concrete places.
Political decisions get made on simplified pictures of the sector. When an MEP, a senator or a regional minister sits down to debate regulation, the figures in their head come from the headlines they read the week before. If those headlines confused adoption with return, exaggerated five-year projections or failed to distinguish generative AI from other kinds of AI, the regulation gets built on wrong premises. The European AI Act is probably the best regulatory text in the world right now, but the national public conversation around its application has been poor.
Business decisions get made on those same headlines too. A board that reads the general-interest press believes the sector is moving faster than it is, that its competitors have what they in fact don't, and that the ROI will arrive sooner than it will. That produces rushed investments, badly calibrated hires, unmet expectations and, in the end, disappointment and retreat. The bubble is inflated, in part, by the coverage.
And individual decisions — the student choosing a degree, the professional reorienting a career, the parent deciding what to give or withhold from a child — get contaminated too. There's needless anguish and ill-founded certainty going around.
The political question
This is personal opinion, but I hold it with eighteen months of a notebook logging pieces. The degradation of media coverage of AI in Spain isn't fixed by individual goodwill. It's fixed by three concrete things, and only those three.
The first is that the press mastheads decide their technology supplements aren't profitable only if subsidised by sector advertising. That forces them to grow the staff, to pay for original reporting, and to accept that certain advertisers will leave when criticised. It's a hard business decision, one that runs through recognising that covering a sector riddled with conflicts of interest is public service, not a revenue stream.
The second is that readers with the capacity to pay subscribe to specialist sources, Spanish and foreign. The market moves when there's solvent demand. If only free, low-quality coverage gets consumed, that's what the market will produce. Subscribing to Stratechery, to Platformer, to 404 Media, or to the few Spanish newsletters doing serious analysis — including, on a minuscule scale, this blog — is the operational way for the ecosystem to hold up.
The third is that readers learn, patiently, to distrust headlines. When a headline sounds too clear or too dramatic, it almost always is. The reader's methodical doubt in the face of the headline is the only defence against coverage that will still take years to improve.
The concrete figure that underwrites the urgency: according to the Digital News Report 2025 from the Reuters Institute (Oxford, June 2025), the share of Spanish readers who pay for online news sits around 13%, against 24% in the United States or 26% in Norway. That gap in figures is what most explains, in a single line, the gap in coverage the reader gets.
Definitions
Press release: a document produced by the company that is the subject of the news, written in the frame and terminology that suit the issuer. Its usefulness to journalism is as a primary source; its direct use without cross-checking is covert advertising.
Original reporting: journalistic work that includes searching for sources who didn't send a release, independently verifying data and building the piece from the ground up, not from a pre-existing text. It's the most expensive genre and the most necessary.
Quick translation: an informal genre of journalistic piece that takes a story already published in another language and carries it into Spanish with minimal rework. It has an operational function but low informational value.
Paying for news: an indicator from the Reuters Institute Digital News Report that measures what share of the population regularly pays — by subscription or per article — for online journalistic information. Spain has stayed below the European average for as long as the indicator has been collected.
References
Reuters Institute for the Study of Journalism, Digital News Report 2025 (Oxford University, June 2025). Source of the figures on subscription and news consumption by country.
Asociación de la Prensa de Madrid, Informe anual de la profesión periodística 2024 (APM, 2024). Figures on staffing, working conditions and the business model of technology sections.
Maldita.es, the Maldita Tecnología section and annual publications on technological disinformation (2022–2025). Sector fact-checking work.
Marta Peirano, El enemigo conoce el sistema (Debate, 2019). Critical analysis of the platform economy that has shaped the best Spanish-language coverage of the sector.
Cory Doctorow, The Internet Con: How to Seize the Means of Computation (Verso, 2023). A frame for understanding why the information ecosystem concentrates and degrades.
Nicholas Lemann, Transaction Man: The Rise of the Deal and the Decline of the American Dream (Farrar, Straus and Giroux, 2019). Background on how the relationship between press, business and power changes when the business model transforms.
Neil Postman, Amusing Ourselves to Death (Viking, 1985). The canonical text on how the medium conditions the content and why certain formats can't hold certain conversations.
Going deeper
Stratechery (Ben Thompson, stratechery.com). A paid newsletter that is the global reference for strategic technology analysis. Essential for anyone wanting to understand the sector beyond general-interest coverage.
404 Media (404media.co). A US journalism cooperative founded in 2023 by former Vice Motherboard reporters. It does investigative reporting on technology.
Platformer (Casey Newton, platformer.news). A newsletter on digital platforms with a focus on moderation and policy.
Newtral (newtral.es). A Spanish outlet with a verification and analysis section on technology and politics.
You might also like
- Stratechery, Platformer and 404 Media, AI journalism done seriously
- OpenAI's press release is not journalism
- How the public conversation about AI broke
- How the press is using the AI Index 2026 figures
Comments0
No comments yet.
Leave a comment