Open Source Community Involvement

While happy for OpenAI’s o3 , I’ve decided to end my OpenAI Pro subscription immediately and move 100% to open-source models like (our own) Hominis and DeepSeek. Here’s why: 1. Transparency Over Opacity: Open-source models allow anyone to inspect, modify, and improve their code. This transparency builds trust, fosters accountability, and ensures there’s no "black box" governing how decisions are made—a critical factor in ethical AI. 2. Community-Driven Innovation: Proprietary models are shaped by corporate priorities, but open-source projects thrive on collaboration. By supporting open-source, I’m investing in collective progress over centralized control, empowering developers worldwide to push boundaries equitably. 3. Customization Without Limits: Closed systems often restrict how tools can be adapted. With open-source, I can tailor models to my specific needs, whether for creativity, research, or problem-solving—without waiting for a corporation’s permission or roadmap. 4. Ethical Independence: Relying on a single company’s AI ecosystem risks amplifying its biases, limitations, or profit-driven motives. Open-source alternatives decentralize power, ensuring technology evolves to serve *people*, not shareholders. 5. Long-Term Sustainability: Subscription models lock users into recurring costs, while open-source projects like DeepSeek prioritize accessibility and user agency. I’d rather support frameworks that democratize AI’s benefits, not gatekeep them. This shift isn’t just about tools—it’s a future where technology belongs to everyone.

Why should NGOs develop open source AI for governments, and give it away for free? At DataGénero - Observatorio, through our project AymurAI, we are doing just that and here’s why it matters: In Latin America, as in many parts of the world, governments urgently need to adopt and innovate with AI. But too often, the default path means handing over sensitive population data to large private companies and locking public services into closed, costly ecosystems. We believe there is another way. By developing open source, non-extractive AI and delivering it for free to public institutions, we enable: ✅ Safe and sovereign use of AI: governments can use AI without compromising citizens' data ✅ Equal access: smaller cities, local courts, public services that can't afford commercial tools can still benefit from quality AI ✅ Transparency and accountability: open code can be audited and improved by the community ✅ A different model: not a business model, but a model of public interest innovation, designed to be replicated and expanded, working side by side with government, the private sector, academia and civic society. And why should the private sector help fund this? Because AI ecosystems thrive when they are open, inclusive, and accountable. Supporting public-interest AI is an opportunity to foster innovation that benefits society as a whole, strengthen public capabilities, and build a more ethical, equitable digital future. Since 2022, we have been deploying our software AymurAI across judicial institutions in Argentina and Costa Rica, showing how gender sensitive, human-rights-based AI can power public services. This is not the usual Silicon Valley playbook. It’s a different path and we think it’s one worth scaling, adapting, and sharing. If you know others exploring similar approaches, or if you can help us spread this experience help us spread the word. We’re eager to collaborate and learn with others working to democratize AI for the public good. And also a big shoutout to the startups that helped us to develop and deploy our tool through the years: collective.ai, Aerolab (and their devs and founders: Julián Ansaldo, Raul Barriga Rubio, Lionel Chamorro, Cecilia Giraudo, Ivan Pojomovsky, Luciano Lapenna, Lucía Wainfeld, Adriana B., Julieta Bertolini) #OpenSourceAI #PublicInterestAI #LatinAmericaandAI #DataJustice #AymurAI The Patrick J. McGovern Foundation Vilas Dhar A+ Alliance UNESCO Prateek Sibal Nick Martin Craig Zelizer Perry Hewitt DataDotOrg

Open source powers the software we all rely on - but if it isn't accessible, we're leaving people out. I'm excited to share a new guide just published on opensource.guide that gives maintainers practical, actionable steps to make their projects usable by everyone. Inside the guide: ✅ How to write an accessibility statement and ACCESSIBILITY .md ✅ Making docs accessible by default (headings, alt text, captions) ✅ Designing accessible interfaces (keyboard support, semantics, color & contrast) ✅ Building accessibility into PR checklists, issue templates, and CI ✅ Leveraging GitHub Copilot for accessibility tasks ✅ Continuous testing — automated and manual The core message: "Nothing about us without us." The most impactful thing you can do is partner with people with disabilities - early and often. You don't have to do everything at once. Start small this week: 🔹 Add an ACCESSIBILITY .md 🔹 Make sure every interactive element is keyboard reachable 🔹 Fix missing form labels 🔹 Add alt text to your README Every fix opens the door for someone who couldn't use your project before. That's a win worth celebrating. 🎉 Read the full guide: https://lnkd.in/eWWwJYJ7

What if the tool you're building is making the web less accessible - not because of how it works, but because of what it lets people create? Most web developers have heard of WCAG (Web Content Accessibility Guidelines). It's the go-to standard for making websites accessible. But WCAG isn't the only accessibility guideline out there. And if you're building anything that lets users create or publish content online, there's another one you should know about: ATAG. ATAG stands for Authoring Tool Accessibility Guidelines. And it applies to more tools than you might think. If you're building a CMS, a blog platform, a social media tool, a website builder, an email editor, a comment system, a forum, or really anything that lets people create and post content, then you're building an authoring tool. That also includes plugins, extensions, and add-ons for platforms like WordPress, Drupal, Shopify, Webflow, etc, or at least the ones that involve content creation. If your plugin gives users a way to input and publish content (think form builders, page builders, review systems, or portfolio tools), ATAG applies to you too. ATAG has two parts, and both matter: • Part A is about making the tool itself accessible. Can someone using a screen reader or keyboard actually use your editor? Can they format text, upload images, and publish without barriers? • Part B is about helping users produce accessible content. Does your tool prompt for alt text when someone uploads an image? Does it flag missing headings or low contrast? Does it make it easy to do the right thing? That second part is where it gets really interesting. Because even if your tool is technically accessible to use, if it doesn't guide users toward creating accessible output, you're potentially multiplying barriers across the web. Think about how many blog posts, product pages, newsletters, and social updates get published every day through tools like these. If the tool doesn't support accessible content creation, that's a lot of inaccessible content going out into the world. WCAG tells us how to make our own sites accessible. ATAG asks us to think bigger - to make sure the things we build help other people create accessibly too. If this is new to you, that's ok. It's not talked about nearly enough. You can find the full guidelines at w3.org/TR/ATAG20, and it's worth a read if you're working on anything in this space. #Accessibility #WCAG #a11y #webdevelopment [Image description: Comparison graphic showing two accessibility guidelines side by side. On the left, WCAG (Web Content Accessibility Guidelines) with the tagline 'Make your site or app accessible.' On the right, ATAG (Authoring Tool Accessibility Guidelines) with the tagline 'Help your users create accessibly.' Each side has an illustration of a developer, working on a website, and pointing at a website as though to guide someone.]

AI coding tools have an accessibility problem. I decided to fix it. I am a screen reader user and accessibility specialist. I use Claude Code every day to build apps at Techopolis LLC. And every day, I have to fight for the fundamentals. Labeled inputs. Focus trapping. Semantic HTML. Contrast ratios. Live regions. These are not advanced requirements. They are the basics. And AI drops them constantly. I tried writing detailed instructions. I tried custom skills. I tried adding reminders to every prompt. None of it stuck. As conversations grow, the model deprioritizes accessibility. Every time. So I built something different. Six specialized AI agents, each with one focused job it cannot ignore. An ARIA Specialist. A Modal Specialist. A Contrast Master. A Keyboard Navigator. A Live Region Controller. And an Accessibility Lead that coordinates them. A hook fires on every prompt. If the task involves UI code, the team activates automatically. If it does not, Claude works normally. It enforces WCAG 2.1 Level AA compliance. It covers VoiceOver, NVDA, and JAWS compatibility. It catches framework-specific pitfalls like React conditional rendering breaking live regions and Tailwind color classes failing contrast. It is open source, MIT licensed, and installs in about thirty seconds. I built it because I need it. And I know I am not the only one. If you work with AI coding tools and care about accessibility, star the repo and share this with your team. The more people involved, the better it gets. GitHub: https://lnkd.in/geYhcZm3 Full writeup: https://lnkd.in/gZdQVxr5 #Accessibility #a11y #OpenSource #WCAG #ClaudeCode #AI #WebDevelopment #AssistiveTechnology #ScreenReader #DevTools #InclusiveDesign

🤗 New from us! Just published a blog post exploring how we're rethinking consent in the AI ecosystem. This comes from my ongoing research into consent mechanisms that go beyond those pesky "I agree" checkboxes we all blindly click (all links in the first comment). Here's what we're seeing in the Hugging Face Hub that differs from traditional closed systems: ⭐️ Community-driven standards: ethical guidelines emerge organically through practical implementation rather than top-down policies. ⭐️ Transparency as accountability: open development processes allow public scrutiny of consent mechanisms that remain hidden in proprietary systems. ⭐️ Diverse implementations: from retroactive opt-out systems to privacy-by-design principles, different approaches tailored to specific contexts. ⭐️ Consent as infrastructure: the most promising systems embed privacy considerations from the earliest stages rather than as afterthoughts. Take Yacine Jernite's Space Privacy Analyzer tool: it uses AI to automatically review Spaces code and generate privacy summaries, helping users understand exactly how their data is handled without wading through dense terms of service/docs -- genius! What's particularly fascinating is how consent in open ecosystems moves beyond legal compliance toward collaborative ethical frameworks. When consent mechanisms develop in the open, they evolve through community experimentation and feedback loops that closed systems simply can't match. The big takeaway? Effective consent isn't about perfect policies; it's about architectures that empower users while enabling responsible innovation. 🚀

𝗧𝗵𝗲 𝗜𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝗰𝗲 𝗼𝗳 𝗢𝗽𝗲𝗻-𝗦𝗼𝘂𝗿𝗰𝗶𝗻𝗴 𝗔𝗜 𝗮𝗻𝗱 𝗟𝗟𝗠 𝗠𝗼𝗱𝗲𝗹𝘀: 𝗕𝗿𝗶𝗱𝗴𝗶𝗻𝗴 𝘁𝗵𝗲 𝗗𝗶𝘃𝗶𝗱𝗲 The rapid advancement of artificial intelligence (AI) and large language models (LLMs) holds transformative potential for numerous sectors, yet it also risks widening existing societal inequities. Historically, technological advancements have often concentrated power and resources in the hands of a few, marginalizing underrepresented groups such as women, people of color, and those from lower economic backgrounds. 𝗛𝗶𝘀𝘁𝗼𝗿𝗶𝗰𝗮𝗹 𝗣𝗿𝗲𝗰𝗲𝗱𝗲𝗻𝘁: 𝗧𝗵𝗲 𝗗𝗲𝗺𝗼𝗰𝗿𝗮𝘁𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗼𝗳 𝘁𝗵𝗲 𝗜𝗻𝘁𝗲𝗿𝗻𝗲𝘁 A compelling historical example of how open-sourcing can democratize technology and foster inclusivity is the evolution of the internet. Here's a brief timeline of its transformation: • 1960s: The Advanced Research Projects Agency Network (ARPANET) develops the internet as a defense project, exclusive to government and academic researchers. • 1980s: The Internet Protocol (IP) is adopted, enabling different networks to communicate with each other and laying the groundwork for a unified global network. • 1991: The World Wide Web (WWW) is invented, making it easy for non-technical users to access and share information using web browsers and hyperlinks. • 1993: The first graphical web browser, Mosaic, is released, popularizing the internet among the general public. This shift was facilitated by governmental and academic initiatives that prioritized accessibility and public utility over proprietary control. Today, we're at a similar crossroads with AI and LLMs. The 𝗢𝗽𝗲𝗻 𝗦𝗼𝘂𝗿𝗰𝗲 𝗜𝗻𝗶𝘁𝗶𝗮𝘁𝗶𝘃𝗲 (OSI) has taken an important step in this direction by defining open-source AI standards. Their initiative emphasizes the need for transparency, accessibility, and collaboration in AI development. By advocating for open-source principles, we can help prevent the concentration of technological power and ensure a more equitable distribution of its benefits. Open-source AI models can help mitigate biases in AI algorithms that often arise from non-representative training data or homogenous development teams. By ensuring that AI systems are freely usable, modifiable, and shareable, we can create a more inclusive and equitable technological future. The push toward open-sourcing AI is about more than innovation or efficiency. It's about creating a future where technology serves all of humanity. Let's use our collective knowledge to build a more equitable technological landscape. What are your thoughts on open-sourcing AI? Share your insights below! 🌍💡

MiniMax Launches Open-Source Model: Accelerating Innovation and Cutting Costs 🤔 Are low-cost, open-source AI models the future, or do premium proprietary systems still have their place? MiniMax recently launched an open-source model, offering free access to its code, models, and weights - the core components that determine performance. Unlike proprietary systems controlled by companies like OpenAI or Google, open-source AI empowers anyone to use, modify, or improve these tools, accelerating innovation and reducing costs. But does this accessibility come with trade-offs? Can these models match the precision and reliability of giants like OpenAI? What about safety and ethics when such tools are widely available? As someone passionate about building responsible AI, I believe accessibility must go hand-in-hand with quality, integrity, and safeguards. Here’s what that means: 1️⃣ Quality: AI must be accurate, consistent, and reliable. For instance, legal or medical AI tools require rigorous testing and regular updates to avoid costly mistakes. Transparency about their limitations is critical. 2️⃣ Integrity: Fairness and trust are key. AI can unintentionally amplify biases in its training data, like favoring specific demographics in hiring tools. Regular audits and a clear process for addressing biases are essential. 3️⃣ Safeguards: Widely accessible AI can be misused. For example, creating deepfakes or phishing emails. Safeguards like watermarking AI-generated content and strict usage policies help prevent harmful applications. Why does this matter? Take MiniMax’s open-source model: it’s a powerful tool for small businesses and researchers, enabling everything from ad campaigns to solving real-world problems. But without safeguards, it could also be exploited for harmful purposes, like spreading misinformation. Open-source AI can truly democratize innovation, but only if we balance its potential with clear rules, responsible practices, and accountability. 👉 So, what do you think? Is the future of AI about low-cost accessibility, or do we still need the refinement of proprietary models? I’d love to hear your thoughts! 👇 Link to the full announcement in the first comment.