Over 650 AI leaders, researchers, and companies — including Meta, Intel, Hugging Face, and AMD — have signed a joint statement titled "Open Source AI Must Win," published at opensourceaimustwin.com, warning policymakers that regulations built for closed, proprietary AI systems could inadvertently entrench a handful of large tech companies.
A Coalition Against Closed Ecosystems
Signatories include Clément Delangue, CEO of Hugging Face; Soumith Chintala, co-creator of PyTorch at Meta; Aravind Srinivas, CEO of Perplexity; and executives from AMD, Databricks, Mistral AI, and Vercel. Their central argument is that closed, proprietary AI systems pose greater risks than open ones because they lack transparency and independent accountability.
The group contends that a future dominated by a few large closed-model providers would concentrate power, limit consumer choice, and slow innovation. They frame the statement as a preemptive response to potential legislation that could impose compliance burdens heavy enough to favor incumbents with the resources to meet them, while squeezing out open-source projects.
The Case for Open Innovation
The statement lays out several arguments for why openness benefits the AI ecosystem:
- Competition and Innovation: Open models let anyone build on, inspect, and improve the technology, fostering a broader market of applications and services.
- Safety and Security: The signatories argue that openness improves security through transparency, allowing a global community of researchers to identify and fix flaws — a claim frequently made in open-source software debates generally.
- Transparency and Accountability: Open models can be independently scrutinized for bias and limitations, unlike closed systems controlled by a single company.
- Economic Empowerment: Open-source AI gives startups and developers a foundation to build on without depending on a single provider's API.
A Direct Appeal to Policymakers
The statement urges regulators to avoid rules that treat the entire open-source ecosystem the same as a handful of powerful frontier labs. The signatories argue that licensing regimes or liability frameworks designed for large closed models would create barriers to entry that crush competition without meaningfully improving safety. Instead, they call for a more targeted regulatory approach that addresses specific, demonstrable harms while preserving the ability to build and distribute open models.
The statement has drawn discussion on Hacker News, where commenters have debated both the sincerity of the coalition — given that some signatories work at companies with commercial interests in open-weight models — and the practicality of its policy asks.