From Geneva, in the middle of the UN Global Dialogue on AI Governance, Chilean Minister Ximena Lincolao took a stance on a dispute brewing in her own country: “none of our initiatives contemplates including Article 8,” the provision that would have exempted artificial intelligence platforms from paying to use protected content (books, music, news, art) to train their models, as reported by Grupo de Diarios América. The statement did not come on just any day: it coincided with the expiration of the deadline for amendments in the Senate Finance Committee for the sweeping reform bill that includes, among other things, that exemption.
The detail matters because it defines who bears the cost of training an AI. “Article 8” would have allowed a model to learn from copyrighted works without compensating their creators, on the argument of enabling a local artificial intelligence ecosystem. Dropping it tips the balance toward authors, but it raises the underlying question: can a small country build its own AI capabilities if it must also pay licenses for all the training material? It is a dilemma no government in the region has yet resolved, which is why the Chilean case works as a trial run watched from outside. That the minister chose a multilateral forum to settle a domestic debate, on the very day of her legislative deadline, shows how much global and local policy intersect today when it comes to AI.
Also today
- OpenAI releases GPT-5.6 for global availability — Its flagship model leaves a preview with access restricted by the US government and reaches the whole world; it is the second frontier lab in a month to emerge from a restriction of that kind, after Anthropic.
- Nvidia and semiconductors lose nearly $1 trillion in two months — The first serious market correction that casts doubt on the narrative of an unstoppable AI “bubble.”
- Cohere buys Reliant AI and reports a rise in demand — The idea of “sovereign AI” gains traction as governments look for alternatives to labs under greater regulatory scrutiny.
- Zuckerberg admits Meta’s AI agents “did not accelerate as expected” — An unusually frank acknowledgment at an internal meeting, after laying off 8,000 employees and promising up to $145 billion in infrastructure this year.
- Joshua Achiam leaves OpenAI — The company’s head of foresight is leaving after almost nine years; it is OpenAI’s fifth senior safety or policy departure in two years.
In the region
The aftermath of the first UN Global Dialogue on AI Governance in Geneva (July 6 and 7) is still moving pieces. The UN and the International Telecommunication Union set up the new 44-member “AI for Good Global Commission,” co-chaired by Rwandan President Paul Kagame and Salesforce CEO Marc Benioff. For the first time it includes the top executives of the big labs as formal members, not as observers, with the stated goal of closing the AI access gap for the 2.2 billion people without reliable internet. No Latin American representation on the commission was confirmed, a gap that contrasts with the leading role Chile is taking today in the debate over AI and copyright.
Launches
- GPT-5.6 and GPT-Live — OpenAI’s flagship model arrives in ChatGPT, its programming tool and the API, and adds a voice feature that lets you speak and listen at the same time.
- Grok 4.5 — xAI’s new model, co-developed with Cursor, with aggressive pricing ($2 and $6 per million tokens); a possible cheaper programming option for teams with tight dollar budgets.
- Muse Image — Meta’s image generator, free within WhatsApp and Instagram; it reaches end users of two apps with very high penetration in Latin America directly, without going through developers.
Threads we’re following
This adds to a story we have been following: that of a region watching global AI governance more from the sidewalk than from the table. First it was the allocation of seats at the Geneva dialogue; now, with two of the three frontier labs most cited in the region (Anthropic and OpenAI) operating under US government approval lists before reaching the global market, the question of Latin America’s capacity to exert influence comes back, this time from the angle of access to models.
If Chile ends up requiring AI to pay for the content it is trained on, will it be a brake on developing local models or the first step toward fairer rules for those who create what machines learn?
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.