[Date TBD!] Safety as the Path of Least Resistance: A New Shape for AI Systems

​After three years of red-teaming frontier models for Anthropic, OpenAI, and METR, Quentin Feuillade--Montixi came away convinced that making the model itself safe is the wrong layer to bet on. Redlines get bypassed, and training hard for them flattens the model into one rigid persona. The leverage, he argues, is in the system around the model.

​In this talk, Quentin will walk through Weft, a programming language for orchestrating AI systems, and the bet behind it: tasks can be broken into scoped steps run by humans, tools, and narrow models, with volition held at the system level instead of in one open-ended agent. The result could be systems that are safer, cheaper, and faster to build.

​He’ll show where this approach already works, where it breaks, and why he believes it could make both regulation and safe deployment more tractable.

7:00pm - doors open, light refreshments served
7:30pm - talk begins
8:30pm - continues hangouts

​Mox Summer Season of events → moxsf.com/summerseason