Free Month in Portugal to Think About AI Safety

WHAT HAPPENED: A group of AI safety researchers just opened applications for the Winter 2027 AFFINE Alignment Seminar. It's a full-time, four-week residency in Lagos, Portugal, from January 4 to 29, 2027. Organizers cover accommodation, travel and food, charge nothing to attend, and hand each of the 35 accepted participants a $1,000 stipend. The deadline is November 22, and they say applying earlier helps.

WHY YOU CARE: AI safety — the work of making sure powerful AI systems do what humans actually want — is one of the fastest-growing research areas in tech, and the big labs are hiring heavily in it. This seminar is unusual because it pays you to think rather than to publish. Mentors include well-known researchers such as Abram Demski and Kaarel Hänni. If you or someone you know has a math background and a nagging worry about where AI is heading, this is a rare free on-ramp into a field with real career momentum.

THE CATCH, HONESTLY: This is not for everyone. You need "solid mathematical footing" and genuine philosophical appetite — the organizers explicitly say the core questions aren't computer science problems. There are no assigned projects and no paper to show at the end, which means no tidy credential for your resume. It's full-time, so you'd need a month free. And with 35 spots and free everything, expect fierce competition. There's also a two-week online reading group beforehand, so accepted participants start working before they fly.

THE BIGGER PICTURE: Many people now think the hardest part of AI isn't building it, but steering it. This seminar bets that the missing ingredient isn't another tweak to existing methods but more people with a deep, holistic understanding of the problem. Whether or not you apply, its existence signals where money and talent in AI are quietly shifting: away from just shipping models and toward figuring out what happens when they get very capable.