Very fascinating paper from Google DeepMind researchers prompts ChatGPT to spit out training data in response to basic queries. This includes large amounts of PII, passages of text pulled from scraped directly from CNN, Goodreads, WordPress blogs, fandom wikis, Terms of Service agreements, Stack Overflow source code, copyrighted legal disclaimers, Wikipedia pages, a casino wholesaling website, news blogs, and random internet comments.
Add comment