AI expert warns against latest trend using ChatGPT to turn photos into caricatures

An artificial intelligence (AI) expert has warned against a viral trend that uses ChatGPT and AI tools to turn personal photos into humorous caricatures, cautioning that the seemingly harmless trend carries serious privacy and security risks.
The trend, which has recently swept across social media platforms, goes beyond altering facial features to create cartoon-like images. In many cases, it attempts to predict a user’s profession and visually integrate it into a satirical scene.

For all the latest headlines, follow our Google News channel online or via the app.

But experts say the process behind these images involves far more than simple image manipulation.
Speaking to Al Arabiya, Mohamed Mohsen Ramadan, head of the Artificial Intelligence and Cybersecurity Unit at the Arab Center for Research and Studies, said the trend conceals real technical and security risks despite its widespread popularity.

He explained that such applications rely on advanced AI systems, including computer vision models that analyze facial features, age, expressions, clothing, and background details. They also use deep learning and multimodal models that combine images, text, prior data, and user behavior.

Detailed algorithmic profiling

Ramadan said uploading a photo does not simply result in a caricature. Instead, the image is subjected to detailed algorithmic analysis to extract traits that may indicate a person’s profession, such as doctor, engineer, journalist, or programmer, as well as lifestyle patterns, professional environment, and in some cases indirect indicators of social class.

He noted that AI does not identify professions in a human sense, but rather predicts them statistically based on facial characteristics, expressions, clothing, image background, metadata, and sometimes links to the user’s social media accounts.

While the results may appear entertaining, Ramadan warned that the process amounts to automated classification that can be inaccurate or biased and often occurs without informed user consent.

Identity theft risks

He added that biometric data, which is considered highly sensitive, can be exploited for deepfake creation, identity theft, and the construction of a permanent digital profile. Linking an image to a digital identity through social media accounts effectively connects a face to a name and online behavior.

When images are used to train AI models, Ramadan said they can become part of vast datasets used to improve facial recognition systems, potentially remaining embedded within models even after an account or application is deleted.

Users, he warned, have no clear way of knowing what ultimately happens to their images.

Data breaches

He also pointed to a more dangerous scenario involving data breaches, noting that leaked images containing detailed biometric features could later be reused for fraud, blackmail, or impersonation, making the data impossible to fully retrieve.
Ramadan stressed that any digital database is vulnerable to hacking.

For his part, Aboubakr Abdel Karim, former assistant to Egypt’s interior minister for relations and media, told Al Arabiya that personal privacy is increasingly under threat.

He said the main danger lies in image retention, noting that photos are not always deleted immediately after processing. Instead, they may be temporarily stored, used to train AI models, or shared with third parties under vague privacy policies.

Abdel Karim urged greater cybersecurity awareness, stressing that the problem lies not with artificial intelligence itself but with its uninformed use.

He advised users to read privacy policies carefully, avoid uploading high-resolution images, refrain from linking such applications to primary social media accounts, and question whether a moment of amusement is worth the potential cost.

Read more:

Elon Musk’s xAI may lose lawsuit accusing OpenAI of stealing trade secrets, judge signals

‘AI surpasses humans in many tasks’: MIT professor shares insights