This ChatGPT Flaw Could Have Exposed Gmail Data Of Users, All Details Here

ChatGPT could be dubbed one of the most commonly used AI models these days, all over the globe. And most of the users have their accounts linked with Gmail. Though patched now, a security flaw was identified in the model that could have let hackers get their hands on users' Gmail data. According to Radware, a popular cybersecurity firm, the flaw was found in ChatGPT's Deep Research agent, which is a tool deployed to help users analyse large chunks of information. And the flaw could have allowed the hackers to siphon sensitive data from both personal and corporate Gmail accounts of the users. A sigh of relief is that Radware said no instances of exploitation were found before the patching process. Furthermore, OpenAI has already confirmed to Radware that the flaw has been

patched. The flaw was so serious that Radware'd Director Pascal Geenes said, "If a corporate account was compromised, the company wouldn't even know information was leaving."

ChatGPT Flaw Could Have Exposed Gmail Data of Users

OpenAI's Deep Reasearch tool available in ChatGPT is built to run online research on the user's behalf, offering quick answers to complex questions. In addition, the tool can also access Gmail accounts with the users' permission. The feature is live for ChatGPT subscribers at an additional fee. Now, with this flaw, the AI agents can be directly manipulated to leak user data without the hacker attempting to attack.Also Read: OpenAI's New Mode Will Let You Decide How Fast ChatGPT Thinks, All You Need To KnowAt the time of demonstration, the researchers sent themselves an email containing hidden instructions that directed the Deep Research agent to scan inbox for personal details including name and addresses. Following that, the AI Agent was prompted to share all this data with a web address controlled by the researchers.