ChatGPT location tracking and deepfake capabilities raise privacy concerns
Malwarebytes testing revealed ChatGPT can identify geographical locations from images without metadata, while a German study found deepfake videos can now convincingly replicate heartbeat signals, making detection more difficult.
What happened
Malwarebytes conducted testing that revealed ChatGPT can determine geographical locations from images lacking metadata such as location data, timestamps, or dates.
A study by Fraunhofer Heinrich Hertz Institute and Humboldt University in Berlin showed that AI can generate deepfake videos containing realistic heartbeat signals, making detection through heartbeat analysis no longer effective.
Reported impact
Malwarebytes warned that ChatGPT can identify locations based on simple visual clues in image backgrounds, such as branded hand-carts or birds with known habitats, raising serious concerns about user privacy.
The advancement in deepfake technology means that current heartbeat detection techniques are no longer reliable for identifying manipulated video content.
Index analysis
Malwarebytes stated ChatGPT appears suspiciously capable of guessing geographical locations by analyzing architectural details, surrounding environment, and other subtle visual evidence in photographs.
Researchers noted that spatial analysis of blood flow distribution could potentially help detect high-quality deepfake videos, offering new methods for detection in the future.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.