Patrun, Inc

Coming off an amazing week at the #AIRISKSUMMIT, where our Founder & CEO Alison Cossette unveiled "Pattern Rights: An Ethical Framework for Generative AI" for the first time. This groundbreaking paradigm -- which led her to found Patrun -- provides a systematic approach to developing generative AI with long-term viability in mind. Placing the human at the center, Pattern Rights offers a framework for understanding the current legal battles surrounding generative AI. It covers: 📖 Copyright 🏠 Data Ownership 👀 Privacy ⚖ Fairness 🌤 Transparency 🇲🇼 Accountability Every innovation cycle brings disruption and a response. #GenAI is no different, though the ethical questions keep getting more complex. 🧠 What's top of mind today: "Risk" thinking forces you to step outside the norm and imagine what might happen. 👉 It's out-of-the-box, requiring systems thinking around how AI and different actors interact in novel ways. We need more of this in the AI ecosystem to foster desired outcomes, and we can learn it from our colleagues in #riskmanagement. #generativeAI #patrun #copyright #AI #intellectualproperty #privacy #fairness #usecase #systemsthinking #ethicalAI

This is exactly why Patrun was created. As AI professionals we knew we had to do better. We were born in September and made our public debut last week at the World Summit AI Americas conference. We are long time fans of Dr. Joy Buolamwini and the The Algorithmic Justice League. We are honored to be part of an AI Community that supports human dignity of creative labor in AI.

Does Your Policy Align with Your LLM? 🤔 We all know AI ethics is a hot topic these days. But have you checked how well your organization's AI policy aligns with the large language models (LLMs) you're using? Many policies emphasize transparency, explainability, and respect for human dignity. However, there could be tensions with popular commercial LLMs like ChatGPT. These models are often opaque "black boxes," potentially conflicting with policies promoting trust in AI. They're also frequently trained on web-scraped data, which could raise copyright and consent issues. I'm not here to call anyone out - we're all learning! But it's important to have open conversations about potential gaps between our policies and the reality of the AI tools we use. So, let's discuss! If you've navigated these challenges or have insights, please share. Together, we can work towards aligning our policies and practices for responsible, ethical AI. Who's in for this dialogue? Let's get it going! 🙌

Hugging Face is hosting datasets that facilitate copyright laundering. The cosmopedia dataset is 30 million text files usable for generative AI training, released under a permissive license that allows commercial use. But the text in the dataset was generated using another model, Mixtral, which we can assume was itself trained on huge amounts of copyrighted work without permission. Cosmopedia is ‘synthetic data’ - data made by another model. It’s generally thought that synthetic data isn’t copyrightable. Because of this, gen AI companies’ lawyers are strongly advising them to use synthetic data - it lowers the risk they’ll be successfully sued. But it’s clearly no more than copyright laundering. You don’t want to train directly on copyrighted work for fear of being sued, so you train on text that was created by a model that itself was trained on copyrighted work. You launder the copyright. This is something creators and legislators need to take very seriously. Some gen AI companies and their supporters explicitly aim to move towards using more synthetic data. This cannot be allowed to be a way to circumvent copyright law. The effect on creators - building a system that competes with them by using their work without permission - is the same.

https://lnkd.in/epKrxgFF

https://lnkd.in/ey4igz-6

We are thrilled to share that we have been accepted as one of 5 startups to present at World Summit AI Americas conference in Montreal in April. The team has been working diligently in stealth mode for months on our mission to bring content creators and AI models into alignment with our new licensing platform. We welcome content owners, customers and the curious to visit us at www.patrun.ai and join our waitlist. Over the next few weeks we will be introducing you to our team, our values and our platform. We welcome you on our journey and look forward a future of ethically sourced models.

https://lnkd.in/eV5UqApm