ChatGPT’s ’80s Movies’ Trend: Information Risks and Environmental Concerns Behind Entertainment

ChatGPT’s ’80s photo’ trend has been spreading across social media in the past few days. Users are uploading their photos to ChatGPT and transforming them into 1980s style. With big hair, period clothing, warm studio lighting, and the grainy feel of old films, these photos look very realistic. However, it is important to keep some privacy issues in mind before participating in this entertainment trend.

What information goes with a photo? A photo doesn’t just contain a face. The photo file can also contain metadata such as the date it was taken, the device the photo was taken on, and in some cases, location information. The background of the photo can also reveal the house number, street name, car license plate, workplace, or the identity of another person.

When images are uploaded to ChatGPT, they are processed by the service as part of the user’s conversation. In a consumer-level ChatGPT account, they can be used to develop content models according to the user’s settings. However, not every uploaded image is automatically used in training; this depends on the type of service and the user’s data settings.

How to reduce the risk? This usually doesn’t pose a major risk for those who are uploading ordinary photos of themselves. However, uploading a photo means giving the platform a copy of the photo that is stored exclusively on the device.

To increase privacy, you can turn off the ‘Improve the model for everyone’ option in ChatGPT’s Settings > Data Controls. Another option is to use ‘Temporary Chat’. According to OpenAI, Temporary Chat is not used in model training, does not appear in chat history, and is deleted from the system after 30 days, with some exceptions.

Users should avoid uploading photos of ID cards, personal documents, or highly sensitive or intimate images. It is best not to upload photos of others, especially children, without permission. It is also a good practice to cut out unnecessary backgrounds, blur house numbers or car license plates, and check privacy settings before uploading photos.

Impact on the environment In addition to privacy, the issue of electricity consumption has also come to the fore in creating images with AI. According to research by Carnegie Mellon University and ‘Hugging Face’, creating an AI image can require an estimated 0.01 to 0.29 kilowatt-hours of electricity, depending on the system and process.

While this may seem small for a single image, if millions of people are participating in the same trend and each creating multiple images, the total energy consumption can be significant. Therefore, creating only the necessary images and not creating multiple versions of the same image over and over again may be a relatively responsible use.