Understand this story
On October 3, 2026, Nathan Lambert and Tom Zick founded the nonprofit organization Trillium Labs, which conducts high-risk AI research publicly. According to Wired AI, the organization launched that day to study directions such as recursive self-improvement (RSI) and agents in a more transparent manner, and to publicly release experimental details for external scientists to reproduce; it initially focuses on post-training while also studying how reinforcement learning shapes model personality and behavior. On funding, it has received undisclosed amounts from Schmidt Sciences, Halcyon Futures and others, plans to raise a total of $40 million to $100 million, and will invest $30 million in training over the next 18 months. On the same day, Nathan Lambert shared links to the team's Wired report, introductory blog, website and recruitment form, along with a sign-up form to join the team.
AI-generated from public reports
Latest development · Report
English translation is being prepared.Developments
2 confirmed developments
- Source:Nathan Lambert
- Source:Wired AI
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Oct 3
natolambert
English translation is being prepared.
natolambert
English translation is being prepared.