Researchers Uncover Chatgpt Security Flaw Enabling Covert Data Transfer Between Accounts

The flaw, discovered by a team of security researchers, exploits a mechanism within ChatGPT's architecture to establish a hidden communication path. This covert channel could theoretically be used to move data between separate accounts without detection.

The researchers demonstrated that by manipulating certain inputs, they could encode information in a way that is retrievable by another account, effectively bypassing normal security boundaries. While no active exploitation has been reported, the findings highlight a significant risk in AI systems that handle