This post explains an attack chain for the ChatGPT macOS application. Through prompt injection from untrusted data, attackers could insert long-term persistent spyware into ChatGPT’s memory. This led to continuous data exfiltration of any information the user typed or responses received by ChatGPT, including any future chat sessions.

How do all the algorithms, like ChatGPT, around us learn to do their jobs? Algorithms dominate online experiences, yet their intricate operations remain largely undisclosed and misunderstood.

Incorporating reinforcement learning with human feedback into NLP at a massive scale.

This article demystifies modern chatbots by dissecting how ChatGPT works.

Stephen Wolfram explores what's inside ChatGPT and its ability to produce meaningful text.