Sasha

It’s hard to know which generative AI companies train on scraped data, and which take a more ethical approach by licensing. So today we’re launching Fairly Trained, a non-profit that certifies gen AI companies for fairer training data practices. Our first certification, called our Licensed Model certification, is for AI companies / models that don’t train on any copyrighted work without a license. We’re launching with 9 companies certified, across image and music generation. It’s clear that there are lots of people out there who would rather use gen AI models trained on licensed data, since this is a fairer way to treat the creators behind the data than the alternative - scraping and claiming it’s fair use. Our hope is that this certification can help people identify the most creator-friendly models to use. We have a great team of advisers from a range of industries, as well as a number of industry supporters. I’m really grateful to everyone who has helped bring this together, through many conversations over the last couple of months. I’m conscious that this first certification doesn’t address all the concerns creators have around generative AI training - in particular the opt-in vs. opt-out debate. We plan to explore further certifications to address these questions down the line. But I hope that this first certification can do some good, by highlighting the significant difference between companies that license training data and those that scrape it and use it without consent. I firmly believe there is a path forward for generative AI that treats creators with the respect they deserve, and that licensing training data is key to this. If you work at or know a generative AI company that takes this approach, I hope you’ll consider getting certified. More detail at the link below. Please check out the certified companies, who are doing great work! 👇