A local flaw put ChatGPT’s Mac app at risk
A trusted Mac process could have helped a harmful local script get past checks in ChatGPT’s app, WIRED reports. The flaw was patched. The story does not show that anyone’s account was breached.
What the source reports
WIRED reporters Lily Hay Newman and Matt Burgess describe a flaw in the ChatGPT Mac app. A harmful script on the same computer could use a trusted process to get past checks meant to screen other programs. That access could expose chat records or let the app take actions. OpenAI acknowledged a fix on September 25. This is a report about what an attacker could do, not proof that any user was breached. The source digest draws a practical line: keep local app access and account connections as narrow as the task allows. A safe setup limits what a flaw can reach, but it does not prove every risk is gone.
Original source
- Title
- A Flaw in ChatGPT’s Mac App Could Have Let Hackers Grab Sensitive Data
- Author
- Lily Hay Newman, Matt Burgess
- Publication
- WIRED
- Date
- Friday, October 2, 2026
Also covered by the other newsletter.