The problem
The most serious risks of advanced artificial intelligence are studied and debated almost only in English: of the major international reports, only the summary exists in Spanish. Meanwhile, the Spanish-speaking world is making decisions that will last for years, such as the creation of Spain’s AI Safety Institute or the positions for the UN Global Dialogue in May 2027, and those preparing them can barely find that knowledge in their own language.
What we do
We work along three paths that do not depend on one another:
- Open knowledge. We translate reference knowledge, with human review, and publish it openly, starting with the complete AI Safety Atlas.
- Training and debate. We organize events, refer people to courses in Spanish and work with universities, the press and our network in Buenos Aires, Mexico and Colombia.
- Advocacy. We bring that knowledge to decision-makers: the AI Safety Institute, the Spanish Parliament and Spain’s positions in the European Union and at the UN.
The change we seek
- Until October 2027, with our volunteer team, we build the foundations: an open corpus, events, a website and contributions to public consultations.
- With funding, we bring on staff to take that support into government offices and newsrooms.
- In 2028, the knowledge is used in Spanish and at least one institution builds serious risks into its decisions.
- In the long term, we help reduce the risk of advanced AI causing a catastrophe. Our share of that global outcome cannot be measured, but every step towards it can.
Why we believe it works
There is demand: 13 events with speakers from four countries, 70 people at our first in-person event, a rating of 4.6 out of 5, and one in four attendees coming back. We also know how to translate at scale: the Spanish version of the AI Safety Atlas, more than 600 pages long, will be ready before the end of 2026. And it has to be done well, because machine translation helps but still fails on specialist terminology.
Our track record and next milestones
- February 2026. iaS is founded, driven by a volunteer team.
- February–September 2026. 13 events with speakers from Spain, Argentina, Colombia and Mexico, and a first in-person event with 70 attendees alongside BlueDot Impact in Madrid.
- Late 2026. The AI Safety Atlas, complete in Spanish.
- March 2027. Our website with all the material openly available, and our first study.
- May 2027. A study on Spanish-speaking positions for the UN Global Dialogue.
- October 2027. Our first public review of what we have achieved.
- 2028. The knowledge is used in Spanish and informs the decisions of at least one institution.
Join us. What is decided in the coming years will affect us all, and there is room for everyone: as a reader, translator, speaker or funder.