Two AI researchers are pushing back against the prevailing secrecy of frontier AI labs. Nathan Lambert and Tom Zick have founded Trillium Labs, a nonprofit that will conduct high-stakes AI research—including work on recursive self-improvement and AI agents—in the open. The lab plans to publish experimental details so outside scientists can study and replicate results, a contrast with the limited-access approach of companies like OpenAI and Anthropic.
Lambert argues that closed development reduces scrutiny and slows the scientific method. "The current closed trajectory of frontier AI development is taking us a step backwards," he told WIRED. The lab will initially focus on post-training, reinforcement learning, and recursive self-improvement, an area that has drawn concern over potential loss of human control.
Trillium Labs has raised an undisclosed sum from Schmidt Sciences, Halcyon Futures, and others. The founders aim to raise $40 million to $100 million total and plan to spend $30 million on training over the next 18 months. They hope to add nuance to the debate over whether powerful AI should be kept behind closed doors or studied openly.