Mozilla

The data is in: open-source AI has reached parity with closed models and costs a fraction of the price — but it's still not making it to production at the rate it should. Today we're publishing our inaugural State of Open Source AI report that includes insights from hundreds of developers. The capability debate is settled. The real decision has moved off the model and onto the harness that still makes open hard to deploy. Full data, geopolitical picture, and where open source AI could head next. stateofopensource.ai

Mark Surman writes in Fortune about why the future of AI depends on more than innovation—it depends on participation. Open-source AI can help create a more transparent, accountable, and human-centered technology ecosystem, ensuring AI is shaped by many voices rather than controlled by only a few. fortune.com/2026/06/29/ope…

Open source is powered by people. Join us at the Open Source Breakfast during Day 2 of the RAISE Summit in Paris, where Mozilla, Hugging Face, Grafana, SUSE, ZML, >commit, and RAISE are bringing together developers, founders, and maintainers for live demos, open Q&A, and conversations with the builders behind today's AI and internet infrastructure. See you there. luma.com/raise26-OSSBre…

🚨 AMA alert! Join @raffi on @Reddit next Tuesday for a conversation about the State of Open Source AI, and our survey of nearly 1000 open source developers. More here: reddit.com/r/MachineLearn…

Capability gap: closed. Inference costs: down 50x. Open models now run a third of real-world AI traffic. So why aren't more companies actually shipping it in production? On July 14, we're publishing the State of Open Source AI Report, including the data, the gaps, and what's next. Stay tuned! 👀

We’re proud to support the launch of the Open Source AI Fellowship from the Open Source Initiative (OSI). Open source has a critical role to play in building AI that is transparent, accessible, and accountable. This fellowship will help advance research, policy engagement, and community collaboration around what it means for AI systems to be truly open source. opensource.org/blog/open-sour…

When advanced AI tools are deemed a national security risk, who gets the final say on whether they stay online? Raffi Krikorian talks with Marketplace about Anthropic’s decision to withdraw access to its cybersecurity AI models — and the broader debate over who controls the AI “kill switch.” marketplace.org/episode/2026/0…

AI has a data problem. Most models are trained on massive amounts of scraped internet data, raising concerns about consent, ownership, bias, and who captures the value. Mozilla Data Collective is building an alternative: community-owned datasets, transparent licensing, and fair value exchange for creators. Trust may become one of AI's most valuable resources. siliconangle.com/2026/06/13/moz…

Who gets to decide when an AI model disappears? Raffi Krikorian argues that the real governance challenge isn't the model itself — it's the power to switch it off without clear rules, transparency, or accountability. If AI can help defenders find and fix vulnerabilities, shutting down access may make us less secure, not more. Shoutout Transformer for covering this: transformernews.ai/p/mythos-fable…

@mignano we couldn't agree more :-) Let's talk! cnbc.com/2026/01/27/moz…

For the first time in ~3 years, it feels like the AI table has been flipped over. Yes, the labs and hyperscalers will have the highest chance of resetting it before everyone else, given their vast capital, frontier model, and compute advantages. But there is now a window for a new ecosystem to emerge. A "rebel alliance," which is what we're calling it at @USV. We are excited by open weight models, distributed compute, human-aligned agents, routing, open source harnesses, the orchestration layer, and a lot more. Basically, anything that gives people and enterprises powerful intelligence while maintaining tight incentive alignment, we're into it.

Have a story about AI stacks, agents, local inference, custom hardware, or emerging standards? Share it with the Developers Wilding track. Proposal deadline: Tonight (June 15th, 2 PM CEST)! Don't just build the future—help define it. mozillafoundation.org/en/festival/tr…

"The central question in AI isn’t open versus closed—it’s owning versus renting.” @raffi and Justin Hendrix discuss the AI security “Mythos moment,” the future of open-source infrastructure, the rise of AI agents and MCP, and why policymakers need to think beyond the model layer to the systems and protocols shaping how AI interacts with the world. Tune in here: techpolicy.press/why-the-ai-pol…

Still doing manually what AI agents could automate? Join Mozilla.ai for a free live webinar on June 8 and how teams are using Octonous to eliminate repetitive work and move faster — no code or big implementation project required. Sign up here: luma.com/octonous-june-…

1/2 Canada just did something big: put openness at the heart of its national AI strategy. By backing open source AI, Canada is choosing a future where AI is something citizens can govern, own and benefit from - not something done to them. 2/2 We’re proud to be partnering with organizations like @Mila_Quebec to help make that vision real.

Sovereign AI is often framed as a geopolitical issue. But what if it's really about control, choice, and resilience at every level—from nation-states to enterprises, communities, and individuals? Mozilla.ai CEO John Dickerson explores why sovereignty in AI goes far beyond who builds the models. Check it out here: blog.mozilla.ai/sovereign-ai-c…

Who really controls enterprise AI agents? Join @raffi live on CXO Talk to hear why open-source AI may be the key to independence, what closed platforms risk, and why visibility and control matter. cxotalk.com/episode/mozill…

AI adoption without trust is a losing strategy. Stephanie Mehta from Fast Company interviews Mark Surman on why companies need to focus on employee agency, meaningful AI governance, and accountability if they want people to trust how AI is being used. “The consequences of being untrustworthy and ignoring accountability are through the roof.” The companies that build trust alongside AI innovation will win. fastcompany.com/91543169/mozil…

The AI stack is reorganizing in real time. Raffi Krikorian’s newest newsletter connects everything from Karpathy’s “microgpt” experiment to on-device AI, federated learning, agent harnesses, and Minecraft being used to bypass censorship and preserve press freedom. Read here: newsletter.ownersnotrenters.com/p/the-complete…

What should AI users actually be allowed to see? In an interview with @TheAtlantic CEO @nxthompson , @raffi argues that open-source AI could make systems more understandable, accountable, and trustworthy — while also raising important questions about safety and control. youtube.com/watch?v=uK7Eca…