🌎 What is the global landscape of #AI (de)regulation? 🌎. In our new #FAccT paper, we aim to cut through the hype & analyze emerging AI regulations across 5 early movers: the EU, US, China, Canada & Brazil. Cc @thesezickbeats @kevin_klyman @BKCHarvard Shira Gur-Arieh🧵

Many thanks to @BKCHarvard for their early support & to all the experts who generously shared insights on local legal contexts 🙏. Full paper here ▶️

Next steps?. 🔉Presenting at #FAccT25 & #IAPP25 this summer. 🗺️ Expanding our geographical scope to include new laws like South Korea’s Basic AI Act. 🧭Turning our paper & data viz into a dynamic tool for global AI cooperation . Want to collaborate? Feel free to DM!.

Key finding 4️⃣ Civic engagement remains a blind spot in #AIregulation. There's little data on whether public consultations translate into meaningful legal outcomes—or are merely performative. Are we engaging citizens or just checking boxes? 👀.

Key finding 3️⃣ Not all #AIregulation is equal. Hybrid frameworks—combining ex ante & ex post interventions with technology- & application-based rules—are the most comprehensive. They address societal harms & national security risks before AND after deployment.

Key finding 2️⃣ Innovation vs. regulation is a false dichotomy. China shows it’s possible to enforce mandatory AI safeguards while developing frontier models like #DeepSeek . While regulatory goals differ across regions, it’s clear countries can lead the AI race & regulate it.

Key finding 1️⃣ We must clarify what #AIregulation means. The term is used ambiguously to describe both binding laws AND voluntary industry guidelines. Lines are strategically blurred between hard law & soft law, opening doors to regulatory capture. 🚩AI regulation ≠ AI policy.

To democratize our insights, we collaborated with @HarvardGSD designers to translate dense AI legislation into accessible outputs. Our aim was to ensure all stakeholders—especially from civil society—are well-equipped to navigate this quickly evolving landscape.

We built a taxonomy comparing AI reg in 🇪🇺🇺🇸🇨🇳🇨🇦🇧🇷 along key metrics:. 🔹Tech vs. application rules. 🔹Ex ante vs. ex post interventions. 🔹Horizontal vs. sectoral scope. 🔹Centralized vs. decentralized enforcement.🔹Maturity of digital legal systems. 🔹Stakeholder engagement.

RT @mmitchell_ai: Pierre refused to accept unless it were also extended to Marie -- his wife, but also his coworker and fellow scientist. T….

RT @mmitchell_ai: Marie Curie was the first woman to win the Nobel Prize. This didn't happen because she herself fought for it -- although….

RT @mer__edith: NEW PAPER!. 1. Small AI models often perform better than big in context. 2. Obsession w bigness has bad consequences, from….

RT @random_walker: 📢 AI Snake Oil is out today! I have too many feelings because this book has been five years in the making. In 2019 I gav….

RT @random_walker: The vast majority of research effort seems to be going into improving capability rather than reliability, and I think it….

RT @mmitchell_ai: In keeping with our core value of transparency @huggingface, @IreneSolaiman and @YJernite have put together details on al….

RT @CarrCenter: Join us with the AI & Tech Policy Caucus for "Surveillance Capitalism in the Age of AI" featuring Shoshana Zuboff (@shoshan….

RT @Abebab: "the Cloud now has a greater carbon footprint than the airline industry. A single data center can consume the equivalent electr….

RT @Abebab: Experts testied 5 SOTA AI models (Anthropic’s Claude, Google’s Gemini, OpenAI’s GPT-4, Meta’s Llama 2, & Mistral’s Mixtral) & f….

RT @BertuzLuca: The Commission has opened the first vacancies for technical profiles in the AI Office today. Only one year of experience is….

RT @DeepikaRaman8: Brb loading up on my hot takes for this mixer.