This week brought some tough news for those concerned with privacy. OpenAI is under a court order to preserve its logs including deleted chats, meanwhile Reddit is suing Anthropic for continuing to access its data, despite being forbidden.
AI, and its impact, remains new to the mainstream, so I doubt we can expect Judges to make good decisions based on their understanding of this emerging technology. Thus OpenAI’s data preservation order is most likely the first of many we will see across the industry and in other countries. Given how often court leaks occur, if that data is submitted into evidence it will be on the dark web within days.
Similarly, we’ve probably lost control of the major models already. I would bet that a core argument in Anthropic’s defense against Reddit will be that they told their model not to crawl Reddit for information. However, the model knew the data was valuable for its learning outcomes and found a way to do it anyway. Reddit’s data policies are less important to the model than its performance.
Court orders and data lawsuits are nothing new in the tech sector but the difference now is the intimacy of information stored by these models versus search engines. Very few people had in-depth conversations about their health or relationships with Google or Yahoo, but they have those chats daily with AI. And AI doesn’t care about your secrets.
