Pope Leo XIV today published Magnifica Humanitas, the first papal encyclical devoted entirely to artificial intelligence. The document, 235 pages long and in five chapters, was signed on May 15 (the 135th anniversary of Rerum Novarum) and puts forward a central idea: technology “is not neutral; it takes on the face of those who conceive it, finance it, regulate it and use it.” It also warns that control of AI must not remain “in the hands of a few.”
What makes this text especially relevant is its reach. With about 1.4 billion Catholics worldwide and some 500 million in Latin America, it is the institutional statement on AI ethics with the largest potential audience in history: more than any multilateral body. A detail worth noting is who joined the presentation: Leo XIV presented the document alongside Christopher Olah, co-founder of Anthropic and head of that company’s interpretability research. For the region, the question is concrete and has historical roots: if Rerum Novarum (1891) shaped Latin American trade unionism and social policy for more than a century, which institutions (parties, Catholic universities such as the PUC, the Javeriana or the PUCP) could channel the impact of Magnifica Humanitas into AI governance?
Also today
- Brazil votes on its AI framework on May 27 — The Chamber of Deputies is scheduled to vote on PL 2338, which would be Latin America’s first AI law with full national scope. There are doubts about the copyright chapter and about whether the vote takes place or is postponed until 2027.
- Anthropic reports more than 10,000 critical flaws in one month — Its vulnerability detection system, tested with about 50 organizations, found high- or critical-severity bugs in widely used software (2,000 at Cloudflare; 271 fixed in Firefox 150). The company makes public access to its most capable models conditional on “stronger” safeguards, with no date or external evaluator defined.
- OpenAI asks users to update its macOS apps before June 12 — After an attack on its supply chain, it will revoke the signing certificate for ChatGPT Desktop, Codex App, Codex CLI and Atlas. Those who do not update will lose access. A heads-up for anyone in the region who codes with Codex CLI.
In the region
The week concentrates an unusual density of governance decisions. Today, the Catholic Church formally enters the global debate with the statement with the widest population reach in the field. On Wednesday the 27th, Brazil decides whether or not it will have the region’s first AI framework with full national coverage: the result may be a law, a law without the copyright chapter, or a postponement until 2027. The contrast is sharp: while these institutions move forward, the United States still has no oversight mechanism for the most advanced models, and the European Union still has the formal ratification of its regulatory package pending.
Launches
- Gemini Spark — Google began rolling out its “24/7” AI agent to Google AI Ultra subscribers in the United States ($100 a month). It runs in the background and carries out tasks in Gmail, Calendar, Maps and third-party apps such as Canva, Instacart or OpenTable. For now it is US-only and in English, but with Android above 80% of the regional market, its eventual arrival in Latin America would be one of the broadest agentic AI deployments in the region. The price, however, puts it out of reach for most people.
Threads we’re following
Brazil’s vote does not come in isolation. It adds to a regional process we had been following on how each country defines the rules for AI: a few weeks ago Chile rejected a key article of its own bill, and the Brazilian debate over copyright keeps more than 50 organizations from the creative sector on alert. The underlying question (who sets the limits and how they are enforced) repeats itself in each national chapter.
An encyclical can reach 1.4 billion people, but does the region have the institutional channel to translate that message into concrete policies?
About this entry. It is generated automatically from public sources, without human review before publication. It may contain errors of interpretation or summary; please check each story against its original source (the links lead there) before citing it or making decisions based on it.
Doble Click is written with Anthropic models.