Your ’80s style AI photo is temporary: What happens to images you upload?

The trend of transforming photos into 80s-style images using AI tools has become popular on social media. However, this trend raises concerns about the data privacy implications of uploading images to AI chatbots. A recent report highlights the risks associated with this practice. When users upload a photo to AI apps like ChatGPT or Gemini, they are not just sharing a picture; they are providing a wealth of personal data.

Photos carry metadata that includes details such as the time taken, the device used, and the location where it was taken. Since faces are unique, uploading an image can indeed be considered personal data. AI systems collect various types of information, including the content uploaded and the user's responses, responses to the system, and data stored after the session.

This means that even after the AI-generated image disappears from social feeds, the original photo may still be stored for a long time, depending on the data retention policies of the AI company. For instance, ChatGPT retains chats for up to 30 days, while Gemini keeps chat data for 18 months by default. These retained data can potentially reveal the user's habits, interests, or profession, even without any identifiable information.

To mitigate these risks, users are advised to avoid uploading overly personal images, especially those of children, and refrain from using AI tools that require identity documents. Turning off model training in settings, using temporary or incognito chat modes, and deleting old chats and files can also help minimize the exposure of personal data.

While AI chatbots are not necessarily more dangerous than other apps, their data collection practices can make them more exposed, making it crucial for users to take necessary precautions.