Media Party arrived in Europe for the first time, and it did so with artificial intelligence as the protagonist. On its opening day at La Caixa Forum in Barcelona, more than 200 people attended a series of keynotes that ranged from rebuilding audiences to rewriting the rules of the internet: from 3Cat’s bet on regaining public trust through community-building, to NPR’s use of AI for investigative reporting, to Cloudflare’s warning about the end of the pact that has sustained the web, and the governance model PRISA Media built to scale AI without losing control. A first day that left a shared message: artificial intelligence is no longer optional for media companies — it’s infrastructure they need to make active decisions about.
La Caixa Forum in Barcelona opened its doors today to a new edition of Media Party, the gathering that has for years brought together journalists, technologists and media executives from Latin America, Spain and the United States. This is also the event’s first European edition, after being centered on Latin America until now. The gathering drew more than 200 people to a day that surfaced one idea running through every keynote: artificial intelligence is no longer an abstract promise or threat — it is a force rewriting the economic, editorial and trust rules of the media industry.
Judith Argila (3Cat): communities to win back lost trust
The first keynote was delivered by Judith Argila, Director of Digital Audience Development at 3Cat. Argila argued that a public broadcaster’s mission today means reaching audiences who don’t know it or distrust what it offers, and helping them cross that “funnel” until they become loyal community members, leaning on native content and distribution strategies built for each platform.
A key idea from her talk: social trust is no longer concentrated only in institutions and media brands, but also in independent journalists, creators and individuals. She described AI as a synthesis of collective knowledge embodied in a single, all-powerful voice, which forces public media to rethink their role as trusted references. She closed with numbers from 3Cat’s creator casting process: hundreds of applications and dozens of new talents brought on board.
Teresa Mondría Terol (NPR): AI as an investigative tool
Teresa Mondría Terol, Manager of AI Products and Platforms at NPR, turned the focus to investigative journalism. She walked through concrete cases of AI use in investigations, including analysis of the Epstein Files and the study of long-term trends in political discourse. She stressed that applying these tools responsibly requires understanding the mechanisms behind the technology, not just its outputs, and framed her talk as a set of practical case studies meant to build intuition for when and how to bring AI into an investigative project — from simple scripts to agentic AI systems.
Chema Alonso (Cloudflare): breaking the internet’s bargain to rebuild it
The most talked-about keynote of the day. Chema Alonso, Vice President of International Development at Cloudflare — the company that handles roughly 20% of the world’s HTTP traffic — argued that the implicit bargain that has sustained the web for decades is broken: automated crawling was tolerated because search engines sent audiences back to journalists and publishers. AI strains that balance by summarizing content and grounding its answers in human-made material without sending traffic or economic value back to the people who produced it.
Alonso didn’t call for blocking AI, but for rebuilding the rules of exchange, since robots.txt doesn’t distinguish between crawling for search, model training, or use by autonomous agents. He cited Cloudflare’s own data: nearly 90% of AI crawler requests are for model training or mixed purposes, not pure search. For Alonso, a granular permissions layer isn’t a tax on AI — it’s the mechanism that keeps the well of human content that AI models themselves need from running dry, preventing model degradation.
The argument builds on what Alonso has been saying since he took on his role at Cloudflare less than a year ago: that a third of internet traffic already belongs to bots and AI, that the web is shifting from pages built for humans to APIs and Markdown built for machines, and that the ad-supported model that sustained the internet is breaking down. He has also flagged his concern about social polarization, and maintains that technology alone solves nothing — it depends on human will to use it to improve society.
Olalla Novoa Ojea (PRISA Media): governing AI before AI governs the newsroom
The morning’s final keynote, from Olalla Novoa Ojea, Director of Artificial Intelligence at PRISA Media, played almost like a warning: govern AI before AI governs us. She opened with two figures: 77% of employees paste internal data into AI tools they manage on a personal basis (Ivanti, 2025), and 45% of AI system responses about current events contain significant errors (BBC/EBU, 2025). What’s at stake, she said, is trust: editorial credibility, legal exposure, data leaks and cognitive delegation.
With more than 140 million monthly unique users and a presence in 12 countries, PRISA Media moved from isolated AI pilots to a governance model built on shared principles. For Novoa Ojea, governance isn’t a brake — it’s permission to move faster: defining what’s allowed before scaling up.
The model rests on a biweekly AI Oversight Committee and a bimonthly Executive Committee that sets priorities alongside the CEO. Every initiative is evaluated against three questions — what, how and why — before getting a green light, a process inspired by the “use case cards” framework tied to the EU AI Act. The results so far: 1,081 people trained, 21 approved tools and 8 projects already in implementation. Even so, she acknowledged persistent challenges: shadow IT hasn’t disappeared, training never feels like enough, and new fronts keep emerging, like governing autonomous agents and safely integrating so-called “vibe coding.”
Closing out day one
Between communities rebuilding trust, newsrooms using AI to investigate better, an internet bargain that needs rewriting, and organizations learning to govern technology before it governs them, day one of Media Party Barcelona 2026 left a shared diagnosis: artificial intelligence is no longer up for debate as an option — it’s infrastructure that demands active decisions.
Tomorrow, Tuesday, the event continues with a second day of keynotes, hands-on workshops and lightning talks, which you can check out on our agenda.
There’s also still time to join the hackathon on September 9 at BIT Hábitat, which requires separate registration. You can find all the details here.
