Get the latest tech news

ChatGPT AI Can Be Fooled to Reveal Secrets

Discover how emotional manipulation prompts can bypass AI safety measures, forcing models like ChatGPT to reveal sensitive information.

How It Works Pricing Blog Tools

Get the Android app

Or read this on Hacker News

Read more on:

Photo of ChatGPT

ChatGPT

Photo of secrets

secrets

Related news:

ChatGPT for evil: Fake IT resumes, misinfo, and more

Inside OpenAI’s Plan to Embed ChatGPT Into College Students’ Lives

ChatGPT used for evil: Fake IT worker resumes, misinfo, and cyber-op assist

« Linux 6.16 Introduces New Helper For Restricting Symbols To Select Kernel Modules

White House security staff warned Starlink is a security risk. Starlink satellite connections in the White House bypass controls meant to stop leaks and hacking. »