On the Internet everything is shared with incredible ease, we often do it without stopping to think about the possible consequences. Therefore, for years, privacy experts insist on the same: we should not publish anything that we are not willing to Show publicly. Although it may seem evident, the warning also applies to private accounts. Because what is “only for friends” today, tomorrow can be exposed without prior notice.
That risk has just acquired a new dimension. It no longer depends only on what we show, but also on what technology can deduce on their own. The images we publish can hide valuable information, such as the place where they were taken. And with the arrival of artificial intelligence models capable of analyzing and reasoning with photographs, the exhibition is even greater.
OpenAi’s AI can know where you took a photo. The new OpenAI models, known as O3 and O4-mini, have brought visual reasoning to a new level. They are able to analyze images with surprising precision, and combine that ability with tools such as web search and image editing to further refine their answers.
This allows them, for example, to give you better explanations than those of an instruction manual or help you understand a complex plane. But it also opens the door to uses that should make us reflect.
New viral trend. One of the latest trends in networks such as X has nothing to do with creating Ghibli -style images or Lego -style compositions. Now, many users are using these models to identify the exact place where a photograph was taken, even when it does not include metadata (EXIF data).
Just tell the model that they are playing Geoguesr to start analyzing the image, cutting details, looking for coincidences and reaching a conclusion. In one of our tests, the system managed to identify a concrete street in Madrid from a simple screenshot. It took about 15 minutes, but gave a very precise answer.
A feature that should make us think. In an environment as hyperconnected as the current one, where the photos are constantly shared, we must keep in mind that it is not necessary to explicitly geolocate an image so that others can find out where it was taken.
Artificial intelligence has raised the level of exposure without many realizing. And although this capacity has interesting applications, it also raises important risks. Privacy, more and more, depends not only on what we share, but what others can deduce from it.
Images | Screen capture
In WorldOfSoftware | O4-mini is much more than another model of AI. It is the Tesla Model 3 of OpenAI
