A Researcher's Shocking Discovery: OpenAI's Data Practices

Meet Suchir Balaji, a former employee at OpenAI. He's got some serious concerns about how his old company collects data for AI models like ChatGPT and DALL-E. Balaji claims that OpenAI isn't playing by the rules when it comes to copyright laws. He says OpenAI took data from the internet to train their AI, and this isn't fair use or legal. Imagine if someone took your book, artwork, or even your website content without permission. That's kind of what OpenAI is doing, says Balaji.

The AI models learn from this data and create stuff that isn't exactly the same but also isn't completely new. Balaji thinks this is a big no-no. OpenAI, on the other hand, argues it's all fair and necessary for innovation. But the laws are still unclear, and there are plenty of lawsuits happening to sort this out. Balaji left OpenAI because he couldn't support their data practices anymore. He believes AI should benefit society, not harm it.

Actions