Mozilla CTO outlines an open, user-controlled AI vision
Mozilla CTO Raffi Krikorian told Tim O’Reilly at AI Codecon that users should have personal AI systems that can be trusted with secrets and prioritize their interests over their creators’ business models. He argued that open models need to be paired with open memory and storage protocols, authenticated tool calling, runtime environments, guardrails, and evaluation harnesses, with Mozilla aiming to provide a starting point for that ecosystem. Krikorian also said a well-tuned harness can improve model performance by as much as 20% on some coding tasks.
Mozilla’s initiative targets infrastructure beyond model weights, including memory, tool calling, deployment, guardrails, and user-centered governance for personal AI.
This source provides an excerpt. Continue reading on the original site.