We launched our AI program at HRF in early 2025 Our first move was asking to join us He denied knowing anything about AI but thankfully agreed to join us and has been the perfect leader for this initiative We started to film a bit here and there as we began launching a few initiatives and experiences around helping dissidents embrace AI as a tool for freedom. Towards the end of 2025 and especially in January of this year, the tools grew shockingly good to the point that non-technical people could start creating very powerful things very quickly We didn't know WHAT exactly would happen next, but planned to film as much of our work as possible from January to June and then release a mini-doc about the experience Here it is... and it has exceeded our wildest expectations This short video traces the evolution of a simple idea (that AI should be a tool of freedom) into what is now a powerful product (an interface where the world's top dissidents can use versatile and staggeringly-effective personal agents to superscale their work, and critically, they can do this without the inference provider being able to spy on them) Very thankful for everyone involved especially Justin... and Skyler and Paul and Austin and Brandon (the Finite team, they create the activist learning experience "Agent Camp" as well as the Finite AI product itself)... and Rod Roudi at the AI Freedom Lab (Rod produces the AI Hack for Freedom event series)... and Scott at Cultivator Labs (his team captured and directed this film)... and many others including Jack and Miles and team at Block who are truly showing us what is possible with organizations and AI. And of course all the amazing human rights defenders and allies like the Reynolds Foundation who have journeyed with us on the adventure. It has been fun and honestly exhilarating at times but I feel very confident predicting that the big fun still lies ahead 🤖 🚀

I am not really seeing this. We are able to power a bunch of users on like $4k of our own equipment to do some pretty decent work right now. This would have been impossible 6 months ago. Let’s see where we go from here!

Phish reference 😅

I mean. It is *increasingly possible* to offer high quality open weight access Certainly better than a year ago Even pay as you go See

You Enjoy Myself

Directionally that is the goal, funding local inference For now we are testing TEE with a few dozen people on a beast setup that we will soon upgrade to Kimi 2.7 Meanwhile you can do a surprising amount on a spark or something comparable with Gemma! We are doing that too DM me if you wanna test either

Kimi 2.7 and GLM 5.2 both out in the 24 hours around Fable ban 💪

What an amazing run this has been…

oslofreedomforum.com

Don’t trust, verify