ChatGPT geo-location feature raises doxxing concern

Share:

“ChatGPT’s latest models can now guess image locations with surprising accuracy—prompting both awe and alarm.”

OpenAI’s latest models, o3 and o4-mini, have turned ChatGPT into a surprisingly accurate GeoGuessr-style location identifier, according to a new report from TechCrunch. Users are now widely sharing its impressive ability to determine image locations—a trend that has gone viral on social media.

This feature highlights the platform’s improved visual reasoning and ability to analyze image content, even performing tasks like cropping and rotating. One viral example, shared by AI YouTuber Brendan Jowett, demonstrates the chatbot’s stunning geo-locating accuracy.

However, the trend has sparked concern. “The newly popular ChatGPT ‘reverse image search’ functionality has privacy implications,” Jowett notes, pointing out the potential for doxing—publicly exposing someone’s private location online.

TechCrunch notes this capability isn’t entirely new, but the trend has grown with awareness. Interestingly, GPT-4o, despite not being image-enabled, “can sometimes outpace o3.”

OpenAI responded, saying it “spends time training models to refuse requests for private or sensitive information,” and will act against policy abuse.

READ  MORE AT TOMS HARDWARE

Join Our Community to get Live Updates

Leave a Comment

We would like to keep you updated with special notifications.

×