ChatGPT can now search years of your texts—but security experts are most worried about the people who never opted in
One person can give the AI access to years of conversations without anyone else in those threads granting permission.
ChatGPT, the popular AI-powered chatbot, can now search through years of text messages without users' explicit consent. An Apple Messages plugin has been introduced, allowing ChatGPT to access past iMessage, SMS, and RCS conversations. This new feature lets users summarize group chats, draft replies, and even send messages from their Mac. However, security experts are raising concerns about the potential invasion of privacy and the implications of granting AI access to sensitive digital information.
The plugin operates locally by default on a user's Mac, only reading messages when explicitly requested. OpenAI claims that it does not create an index of users' messages or begin reading conversations without a specific request. Nevertheless, Paul Walsh, a security and privacy expert, warns that the plugin can be likened to spyware, as it functions like a third party silently monitoring private communications.
Once enabled, ChatGPT can access conversations initiated by people who never opted in, potentially violating trust and breaking the fundamental concept of end-to-end encryption.
The plugin's permissions include AppleScript, Accessibility, and Full Disk Access, which raise broader security concerns, such as the potential for hackers, insiders, or government agencies to gain access to sensitive information. Even though the encryption itself remains intact, storing conversation data on third-party servers could create a "side door" for unauthorized access.
Privacy-focused technology company Proton has echoed similar concerns, emphasizing the importance of protecting users' privacy, even for those who do not utilize ChatGPT.
Written by urgent.news from Fortune's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
