Why Block signed the Open Weights letter

Block has signed the Open Weights and American AI Leadership letter alongside NVIDIA, Microsoft, Meta, IBM, and more than 100 other companies. The letter urges US policymakers to support open weight AI models and avoid premature restrictions. It makes the case that openness strengthens competition, expands access, can strengthen security when paired with adequate investment in auditing and maintenance, and ensures organisations can run, adapt, and build on AI independently rather than ceding control to a single provider. We signed because this position reflects how we've approached open source for years, and it's how we think about AI as well.

Open weight models mean the organizations using AI can access the model freely, change it, and run it wherever they need to. Startups and researchers get to build on the best available models without starting from zero or paying per-task fees to a handful of providers. Everyone retains ownership of what they build, and transparency allows the broader community to help identify and address weaknesses.

Open source has been central to how Block builds for more than a decade. We've contributed foundational libraries like OkHttp and gRPC, and recently donated four of our repos to Commonhaus Foundation. In January 2025 we open sourced goose, our model-agnostic AI agent, and in December 2025 we cofounded the Agentic AI Foundation under the Linux Foundation alongside Anthropic and OpenAI, contributing goose as a founding project for open agent standards. This month, we launched Buzz, an open source workspace for humans and agents built on the Nostr protocol, and we continue to invest in open frontier model research with the goal of sharing what we build.

For us, open source is a civic resource and a public good that we believe should be treated as such. We believe control over foundational technologies should be broadly distributed rather than concentrated, and we'll continue to support open source approaches to AI while engaging constructively on reasonable safeguards.