A recently patched vulnerability in ChatGPT's macOS app could have allowed hackers to access sensitive data, including chat logs.

The bug, discovered by Objective-See Foundation researchers, highlights the deep system access AI platforms require, making them prime targets for attackers.

‘Agents need a lot of access to do their job’ explained Patrick Wardle, a software analyst. ‘If they can be corrupted, that’s a big problem, as unprivileged code could gain access to all the data.’

OpenAI acknowledged the flaw and swiftly fixed it, but the incident underscores the need for increased security in AI applications.

‘AI companies are focused on adding features but must not overlook security,’ Wardle warned, noting that every new feature broadens the attack surface, making security an ongoing challenge.