On May 25 Pope Leo XIV published Magnifica Humanitas, his first encyclical, devoted entirely to artificial intelligence. Given its reach, it is probably the statement on AI ethics with the largest potential audience in history: some 1.4 billion Catholics worldwide, about 500 million of them in Latin America. What is notable is that the document does not stop at principles: it lands on concrete policy proposals. You can read the official announcement on Vatican News.
Three ideas will resonate most in the governance debate. First, a call for an international treaty banning lethal autonomous weapons systems, on the premise that no algorithm can make a war morally acceptable. Second, the explicit denunciation of “data colonialism” as a new form of exploitation between the North and the Global South, a framework that hits close to home for the region. And third, a striking proposal: that AI patents with high social impact be declared the heritage of humanity, with guaranteed universal access.
The Latin American angle is not a minor one. The Latin American Episcopal Council (CELAM) reacted the same day, and media outlets and portals across the region (from ACI Prensa to the Jesuit portal CPAL and Página 12) quickly picked up its implications. One question remains open: which institutions in the region (bishops’ conferences, ministries, multilateral bodies) could translate these principles into something operational.
Also today
- OpenAI closes its first content deal with Brazilian media — Folha de S.Paulo and UOL will bring real-time news content to ChatGPT, and the deal also ends the lawsuit Folha had filed in 2025. It is the first agreement of its kind between a Latin American media outlet and a frontier lab; the amount is confidential.
- Brazil regulates AI in its election campaign — The Superior Electoral Court announced five measures ahead of the October presidential election, including a standing commission and a ban on AI providers suggesting or favoring candidates. It is the first nationwide electoral system in the region with explicit rules for AI during an active campaign.
- ChatGPT launches “Trusted Contact” for mental health crises — It lets users designate a trusted person who receives an alert if the system detects a suicidal crisis in the conversation. It is the first mechanism of its kind in a mass-market chatbot that triggers an external action; for now it is reaching selected regions.
- Anthropic moves toward a $30 billion round — According to Bloomberg, the deal would value the company above $900 billion and could close this very week, which would make it the most valuable private startup in the world. To put it in perspective: that figure is equivalent to five times the region’s most ambitious sovereign compute plan.
In the region
Brazil is this week the epicenter of AI governance in Latin America, with two moves in parallel. In addition to the TSE’s electoral rules, the Chamber of Deputies has scheduled the floor vote on PL 2338 for May 27; the bill aims to be the region’s first AI legal framework with full national scope. It is not guaranteed: Senate Vice President Eduardo Gomes warned that the electoral calendar could push the discussion to 2027, while the Speaker of the Chamber, Hugo Motta, is sticking to the date. Tomorrow’s outcome will determine whether the region gets its first comprehensive AI framework now or waits another year and a half.
Launches
- Gemini Spark — Google — Google began delivering its agent that runs 24/7 in the background, integrated with Gmail, Maps, Calendar and third-party apps via MCP (Canva among them). For now it is available only to AI Ultra subscribers ($100 a month) in the United States and in English. With Android above 80% of the market in much of the region, its eventual arrival in Latin America will be a barometer of access to consumer agentic AI.
- Claude Managed Agents: private sandbox and “MCP tunnels” — Anthropic — An enterprise feature in public beta that lets Claude agents run tools inside the customer’s own infrastructure, connecting to internal APIs without exposing the internal network to the internet. It is designed for regulated sectors (banking, health, government) that cannot send data to the public cloud, a common scenario for institutions in the region with data sovereignty requirements.
Threads we’re following
It is worth connecting two of the day’s stories. On one hand, the encyclical proposes treating AI patents with high social impact as a common good and warns about the concentration of power. On the other, one of Anthropic’s co-founders took part in the document’s presentation, the very same week his company is moving toward a round that would value it above $900 billion. The tension is hard to ignore and will probably keep surfacing as these labs grow: the same actor the Church highlights for its work on safety and alignment is also part of the enormous concentration of resources the text questions.
Can the same company be, at the same time, a reference in AI ethics and an example of the concentration of power that this ethics seeks to limit? The answer will say a lot about the decade ahead.
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.