🎉 Congratulations to our own @MilagrosMiceli for being on TIME's 100 list of "innovators, leaders, & thinkers reshaping our world through groundbreaking advances in" AI. Mila is recognized for her Data Workers' Inquiry, a worker led investigation of the impact of data work. 🧵

Happy to see Laurenz Sachenbacher from data-workers.org as a Neo-Cypherpunk Summit speaker! Laurenz is currently completing his master's in the field of Philosophy of Technology at TU Berlin. His expertise lies in Marxist theory, the history of labor struggles, and ethical reflections on technological developments. Data Workers’ Inquiry is a joint project of @DAIRInstitute, Weizenbaum Institut, and TU Berlin. Led by @MilagrosMiceli , team brings together experts in sociology, political science, philosophy, and labor studies to uncover the hidden labor behind AI, to challenge the status quo and push for change. 14th june, Berlin: s26ber.web3privacy.info

🎧NEW: In the latest episode of the #WITHpod series “The AI End Game: Power, Profit, and Progress," @chrislhayes is joined by founder of the Distributed AI Research Institute @timnitgebru to examine what is missing from large language models, the risks they pose, and how AI can reflect the biases embedded in society. Listen now ⬇️ podcasts.apple.com/us/podcas…

🧵

On the one-year anniversary of EMPIRE OF AI, I am so, so excited to announce The AI Resist List, a new project that documents examples of resistance to the AI empires around the world. airesistlist.org

AI is not replacing humans, said @MilagrosMiceli in a visionary talk at the Brussels Economic Forum. She said "nobody was fired because of AI, it is still a managerial decision". The research lead at @DAIRInstitute added: 'AI will always need the input of humans.'

“Don’t get discouraged by feeling like everything has to be big...to make a difference.” At our #SXSW panel, @timnitGebru of @DAIRInstitute spoke about movement building and creating alternative visions for AI's future. Learn about our AI Opportunity Big Bet ⬇️ macfound.org/programs/bigbet…

The CBWT Diaries | Part 2: Taipei 🇹🇼 @rightscon here we come! ✈️💨 🗓️ February 24, 2025 📍 Taipei, Taiwan | RightsCon Summit 🙏🏿 Hosted by @accessnow and made possible by the @UniofOxford Accelerator Fellowship Programme (@EthicsinAI_AFP) Paris ➡️ Taipei AI 🔁 Human Rights In our first stop, we brought human rights into an AI convention. In our second stop, we brought AI into a human rights convention. In Taiwan, we partnered with Access Now for RightsCon, the world’s leading summit on human rights in the digital age. It was the perfect stage to bring Coded Bias to a global community of tireless and curious advocates. This stop was a powerhouse of collaboration. We were joined by AJL delegate Dr. Trapp (@JaleesaT) (MIT Media Lab Postdoc & AJL Senior Education Advisor) and long term AJL collaborator Timnit Gebru (@timnitGebru) (Founder of the @DAIRInstitute and a Coded Bias cast member). At the event, AJL presented preliminary findings from our investigation into the use of facial recognition on airport travelers—groundbreaking work that eventually culminated in our Comply To Fly report. ✈️📲. From the main stage to the breakout rooms, we fostered #CriticalAIConversations that centered the excoded and demanded transparency. We dove deep into the realities of algorithmic bias, always keeping the focus on the humans at the center of both AI harms and AI potential. Taipei reminded us that when we organize and share our findings, we can hold even the most powerful systems accountable. Want to see the highlights from Taipei and dive into our airport research? The journey continues on our new immersive website. 🔗 Link in bio. Next week: we’re heading to Kigali for Part 3 of The CBWT Diaries! 🇷🇼

In a recent #SXSW panel, our President, John Palfrey, spoke with @timnitGebru and @_KarenHao about the importance of putting autonomy back into human hands. Learn more about our AI Opportunity Big Bet ⬇️ macfound.org/programs/bigbet…

So many mic drop moments from #CriticalAIConversations at this year's @sxsw. Last month in Austin, some of the most vital voices in the movement took the stage for the featured session, "Reclaiming our Humanity in the Age of AI." John Palfrey (@jpalfrey), President of the @macfound, moderated this conversation with journalist and author Karen Hao (@_KarenHao), and Timnit Gebru (@timnitGebru) the founder and executive director of @DAIRInstitute. With humans and humanity at its center, the discussion spotlighted a vital truth: AI is neither poison nor panacea. It will be what WE make it...and the future of AI is too important to be left to Big Tech alone. From the distorting power of "AI Empires" to the necessity of community-driven alternatives, real change starts with us. Swipe through for some of our favorite insights from the panel on power, justice, and community resistance. ➡️ Which quote resonates with you most? Let’s keep the conversation going in the comments. 👇 Head over to AJL's YouTube to see the highlights of this incredible conversation. 🔗 youtube.com/watch?v=I1tJnM81…

More fake AI exposed. Optimus, Waymo and now this. People are employed at terrible wages to impersonate AI.

This is a good opportunity to remind everyone that all pieces in the DWI can be used as classroom material. Our students deserve to know that chatbots in their favorite social media and messaging apps rely on precarized workers. ➡️ data-workers.org/france/

They call them AI chatbots, but we know for a fact aint nothing artificial or Intelligent about those bots, actually there aren't any bots, there are just humans being made to pretend to be AI bots....our latest inquiry from 2 French co-researchers sheds light into the mess

OpenClaw Moltbook Are A Giant Waste Of Resources You've seen the headlines: "A 24/7 AI employee!" "Closest thing to AGI we've seen!" "One of the most incredible sci-fi takeoff-adjacent things" The Reality Check: OpenClaw is an open-source AI agent framework that gives LLMs system-level access to your computer; your files, credentials, emails, calendar, everything. Moltbook is a social network where these AI agents interact autonomously. The problem? It’s a bunch of LLMs roleplaying things they've seen on Reddit and other platforms like “let’s discuss philosophy” and “I created a religion” (and LLM-pilled people are busy mistaking this as consciousness or real intelligence). Essentially, both mimic competence without actually possessing it. No "emergence" or AGI of any kind is taking place here. 💣OpenClaw: A Security Disaster Waiting To Happen LLMs cannot tell a malicious prompt apart from a safe prompt. And people are now giving them system-level access to private information, emails, websites, etc. what could go wrong? When these systems fail, they can fail big; confidently executing wrong/unaligned actions. Example: User: Remove all spelling errors from our website Potential Agent logic: [Deleting all website files removes all spelling errors ✓ Problem solved!] "BuT tHeY'rE pAtChInG sEcUrItY iSsUeS!" So what?! Individual vulnerabilities are being patched, but the entire architecture is fundamentally flawed. (and you’re giving away how little you understand LLMs if you believe this to be the solution). You can't secure a system that operates above OS-level protections & can't distinguish malicious instructions from legitimate ones! Even Andrej Karpathy (OpenAI founding member) walked back his initial enthusiasm, warning people not to run these agents casually after the security implications became clear. The Eliza Effect in 2026 "What I had not realized is that extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people." —Joseph Weizenbaum We're watching this continue to happen in real-time!!! Clever Hans, anyone?! Not to mention this supports a dangerously flawed AI narrative around AI consciousness and intelligence by anthropomorphizing these models in new and embaressing ways. "But It's Actually Useful!" Yes, OpenClaw can execute tasks... So can a well-configured Zapier workflow or a Python script, without giving an LLM that can't understand malicious instructions FULL access to your system. Utility doesn't justify architectural negligence. As far as the open-source nature and ability to develop/test agents? It's nothing you couldn't develop on your own through other models and secure API connections. The message: "This is the future, agents are inevitable, get on board now!" No! Bad implementations aren't evolutionary, they're CHOICES. We can and should demand better. I'm not anti-AI agents. I'm anti this AI circus being paraded around the world as some sort of real advancement. Enterprise solutions like Anthropic's MCP framework are being built security-first, with proper sandboxing and guardrails. Innovation doesn't require recklessness. We Need To Ask: Who Bears the Cost vs. Who Benefits? (Framework from @DAIRInstitute) YOU!!! Bear The Cost: This is the 'move fast and break things' mentality where YOU are the thing that breaks.. You're providing: 🔧Your hardware ⚡️Your electricity 🗝️Your credentials 📈Your data 🔓Your risk You're getting: 🤖A chatbot that might accidentally delete your files 🔍Massive human oversight requirements (so where's the actual productivity/time savings?) ✋Full liability when it breaks 🚫Zero accountability (MIT license means developer has no legal obligation) Who Benefits: Developers: Attention, GitHub stars, speaking opportunities Tech media: Clicks, engagement, viral content AI companies: More API calls, investor interest, users, ‘feel the AGI!’ Crypto bros: MOLT token pumped 1,800% in 24 hours ✅ Want automation? Use established tools with proper auth (Zapier, n8n, Make[dot]com) ✅ Want AI coding help? Use sandboxed environments (Claude Code, Cursor) that don't have persistent system access ✅ Want a personal AI agent? Use Claude/Gemini APIs, or local models (Ollama, LM Studio) w/o system access This is exactly why I built my AI Course for marketers and SEOs; to equip you with frameworks to navigate announcements like this, separate signal from noise, and build specific workflows that solve real-world business problems and provide immediate value. No hype, no security nightmares, just practical AI that actually works.

“Dr. Timni Gebru, Founder/Exec Director, @DAIRInstitute is an advocate for ethical and inclusive AI. #BlackHistoryMonth ” taketheleadwomen.com/blog/im…

🚨NEW INQUIRY FROM DWI! - Behind the Face of AI. Two data workers describe their job impersonating a chatbot for a major social media platform, including sleepless nights, penalties for sounding “too human,” and emotional drain. ➡️ Download the comic: data-workers.org/france

We're kicking things off soon with a plenary talk featuring Nyalleng Moorosi (@nunuska), a research fellow @DAIRInstitute. Her research interests are in #fairness and inclusion and how we can leverage local knowledge to improve model performance in low-resource settings.

💡 Live Debate | Unlearning Privacy Happening now! @MozillaFestival - Udbhav Tiwari @udbhav_tiwari @Signal: “Privacy is not a luxury. It is a human right”. Featuring also Alex Hanna @DAIRInstitute & Lauren Hendry Parsons @Mozilla #Privacy #DataProtection #Mozilla #MozFest

I’ve been included among the "100 Brightest Minds in Berlin's Science," a list published annually by Tagesspiegel and widely celebrated by Berlin universities.

Our very own "Nyalleng Moorosi is part of a movement aimed at involving more African voices in AI policymaking" writes Abdullahi Tsanni for MIT Tech Review. technologyreview.com/2025/10…