According to recent research, the ability of AI chatbots to soon reach the threshold of excess, some chatbots can locate a person from selfies and some other chatbots can now create almost un lockable "deepfake" videos.
In a series of tests, Malwarebytes discovered that ChatGPT was very good at geographical locations from deleted photos, which often contained detailed information such as location, time, and date.
According to Malwarebytes, users have good reason to worry about what is displayed in the wallpaper because AI is getting very good at predicting your position, based on the smallest clues.
According to Malwarebytes experts, chatbot can use clues and signals in architecture and environment to narrow down the scope of locations that may be in the photo before accurately determining or very closely following the location in the photo.
According to the virus testing software provider, an old car of a specific brand or a bird with a limited habitat appears in the photo, which is enough to suggest the location of the photographer.
The Fraunhofer Heinrich-Hertz Institute and Humboldt University in Berlin have announced the results of the trial showing that AI has the ability to create sophisticated deepfake videos, making it difficult to detect forgery.
Deepfake is a technology that simulates human facial images, named after the phrase "deep learning" ( machine learning) and "fake".
"The current development of image creation techniques makes it increasingly difficult to detect edited content through intuitive testing," the research team said.
Previously, deepfake could not fake heart rates, it almost disappeared in fake videos.
But now, AI has developed to the point where it can create almost real deepfake, suggesting that heart rate detection techniques are no longer effective. However, some high-tech companies can still recognize real or fake deepfake videos by analyzing subcutaneous blood flow, because when the heart pumps blood, the human vein will change color.
