Welcome back, aspiring cyberwarriors!
Geolocation has always been very important in OSINT, but it can take a lot of time and effort. Analysts usually use methods such as matching shadows to the sun’s position, identifying tree types, and comparing building styles to specific areas. While AI-powered image analysis won’t replace these classic techniques, it definitely speeds up the initial sorting process.
In this article, we will explore the custom GPTs offered by ChatGPT, explain what they are, and demonstrate how to use the GeoGuessr GPT for geolocation purposes. Let’s get rolling!
What is GeoGuessr GPT?
GeoGuessr GPT is designed to simplify the analysis process into a single interaction. You can upload an image in the chat interface and ask where the photograph was taken. The model will then provide its best estimate, along with the reasoning behind that conclusion.
Behind the scenes, the model uses a vast training dataset containing city descriptions, architectural guides, travel photography, and millions of geotagged images. It’s important to note that it does not perform a reverse image search. Instead, it synthesizes visual objects to make a probabilistic guess.
The primary problem it addresses is not only providing a final answer but also offering a framework for reasoning. Even if the model is incorrect about the specific city, its chain of observations often gives a reliable starting point for further investigation.
What are GPTs in ChatGPT?
OpenAI introduced the concept of custom GPTs, which allow users to create tailored versions of ChatGPT without writing any code. A custom GPT serves as a wrapper around the base model, including a specific system prompt, optional knowledge files, and sometimes connections to external APIs.
You can find custom GPTs by visiting the GPT Store within ChatGPT or by following a direct link shared by the creator. It’s important to understand that a custom GPT doesn’t possess any unique vision capabilities beyond what the base GPT models already offer. Instead, it features a crafted prompt that directs the model’s general intelligence toward the specific task. You can achieve similar results by creating your own prompts in a typical ChatGPT conversation.
Get Started with GeoGuessr GPT
To get started with GeoGuessr GPT, we need to sign in to our ChatGPT account. After that, we can click on “Explore GPTs” and search for the one we are interested in.
By clicking on the desired GPT, we can explore its capabilities and learn how to start the conversation.
Everything is very simple. So, let’s upload an image and ask GeoGuessr GPT to guess the location.
My first target image will be a house in Hollywood, Los Angeles, CA, USA.
GeoGuessr GPT successfully identified the main objectives and accurately guessed the county and region.
Good, now let’s make the task harder and provide an image from warzone in Ukraine, specifically Mariupol city, at the beginning of the Russian invasion.
Although the final guess in this example was partly correct, referring to Ukraine, it was not about the Donetsk Oblast. What was truly useful for the researcher from around the globe was that GPT explained the core objective: road signs, and so on. This information provided valuable insights, even when the final guess wasn’t completely accurate.
Summary
In this article, we explored what GeoGuessr GPT is and how to start using it for geolocation. It can be a valuable starting point for OSINT investigations. Although custom GPTs do not possess any unique vision capabilities beyond what the base GPT models already offer, they provide an easy-to-use option for beginning the geolocation process.
If you’re looking to enhance your OSINT skills, consider exploring our OSINT training. If you need assistance in uncovering the truth, don’t hesitate to reach out to us at [email protected], and we’ll conduct a comprehensive OSINT investigation for you.
Source: HackersArise
Source Link: https://hackers-arise.com/open-source-intelligence-osint-ai-powered-image-geo-location/