Asking ChatGPT to Repeat Words ‘Forever’ Is Now a Terms of Service Violation

The article discusses how Google researchers have discovered that ChatGPT, an AI language model, is capable of revealing its training data. This data includes sensitive private information and a variety of text sources such as CNN articles, Goodreads reviews, WordPress blogs, fandom wikis, Terms of Service agreements, Stack Overflow source code, Wikipedia pages, news blogs, and random internet comments.

The researchers' attack on ChatGPT has raised concerns about the potential misuse of such information. The findings highlight the need for improved data privacy measures in AI language models to prevent the exposure of sensitive information.

Key takeaways

Google Researchers have found a way to prompt ChatGPT to reveal its training data.
ChatGPT contains sensitive private information.
ChatGPT uses a variety of sources for its training data, including CNN, Goodreads, WordPress blogs, fandom wikis, Terms of Service agreements, Stack Overflow source code, Wikipedia pages, news blogs, and random internet comments.
The information was revealed by Jason Koebler of 404 Media.

Asking ChatGPT to Repeat Words ‘Forever’ Is Now a Terms of Service Violation

Key takeaways

Discussion (0)