About two years ago, a Reddit user shared a link to a new geolocation tool called GeoSpy.
“I created an AI powered image locating tool that lets people find the location a where a photo was taken no exif or metadata needed,” the user wrote.
At the time, Dan Heinen, who made the post, worked as an AI research engineer at a defense company. GeoSpy was a side project he hacked together with his twin brothers. It quickly generated huge interest in the OSINT community, as well as some media scrutiny.
The free version of GeoSpy that attracted so much attention on Reddit is no longer available (though you can still play the company’s location guessing game). GeoSpy Pro is primarily targeted at law enforcement and the product has evolved to tackle challenging geolocation challenges. On LinkedIn Heinen showcases the product’s ability to identify locations from photos with few, if any, distinctive visual details:
And its ability to identify the location/area of photos taken indoors.
I spoke to Heinen about the origins of GeoSpy, how they’re working with low-context outdoor and indoor photos, the difference between GeoSpy and LLMs for geolocation, and the privacy concerns related to such OSINT tools. What follows is a lightly edited transcript of our conversation.
Upgrade to read the rest.
Become a paying member of Indicator to access all of our content and our monthly members-only workshop. Support independent media while building your skills.
UpgradeA membership gets you:
- Everything we publish, plus archival content, including the Academic Library and our Resources
- Live monthly workshops and access to all recordings
