The Hidden Cost of AI: What’s Really Behind the #Trend80 Nostalgia?
Artificial intelligence is rapidly becoming a staple in our daily routines. We ask it to write, translate, analyze, answer questions, and even edit photos. The recent #Trend80 on Instagram offers a striking illustration of this shift. Users around the world are leveraging AI to transform their portraits into an aesthetic inspired by the 1980s, complete with era-specific hairstyles, clothing, backdrops, and color palettes. While the trend taps into nostalgia, it raises a less visible but critical question: what data are users transmitting to these services to achieve such results?
For Habib Soula, a Software Engineer at Value, this is precisely where part of the problem lies. The use of these tools has become so commonplace that users often fail to consider what happens behind the interface. With a touch of humor, he reminds us that any service offered for free necessarily relies on a specific business model. Furthermore, major AI labs possess vast volumes of data to develop and refine their models.
Soula’s point is not to suggest that all AI usage constitutes a threat. Rather, he urges users to look more closely at what they are accepting when using these services. Privacy settings are not trivial; they determine, depending on the service and enabled features, how certain conversations or content may be stored or utilized.
The Specific Risk of Facial Data
The #Trend80 makes this issue particularly concrete. To obtain an image in the 1980s style, users must submit their face to an AI tool. However, a face is not just another piece of data. It can be used to identify an individual and takes on a heightened significance when it involves photos of children or individuals who have not given explicit consent.
Soula also highlights a common misconception: the belief that deleting an app or an account automatically erases all associated data. These two actions are not necessarily equivalent. Data retention and deletion depend on each service’s specific rules, settings, and privacy policies. In other words, uninstalling an app does not necessarily mean that data already transmitted to the service is deleted at the same time.
The Accumulation of Personal Profiles
Soula shares an example involving his mother, who regularly uses AI. His own usage differs significantly; he estimates that nearly 98% of his interactions with these tools are related to development, with the remainder concerning culture, health, or everyday questions. He recounts that two years ago, his mother asked ChatGPT to analyze her medical results. Today, the existence of memory features or the retention of certain exchanges can, depending on service settings, allow the tool to take into account elements from previous conversations.
For Soula, this example primarily illustrates that users can gradually build a much richer history than they realize. An isolated question about health, a photo, a piece of professional information, or a personal preference can, when pieced together, paint a much more precise profile of the user.
Security and Regulatory Concerns
This accumulation also raises security questions. The more data users entrust to platforms, the more strategic the protection of that data becomes. The risk is not limited to how a company decides to process or use the information entrusted to it; it also concerns the potential exposure of that data in the event of a security breach or cyberattack.
Recent incidents in the AI sector have fueled this debate. Soula specifically mentions alerts raised by certain researchers and former employees of major industry players regarding the risks associated with the rapid development of these technologies. He does not advocate abandoning AI. Instead, his argument centers on the necessity of maintaining control mechanisms and a regulatory framework as model capabilities continue to advance.
A Broader Digital Reality
This vigilance is not limited to ChatGPT, Gemini, or other generative AI tools. The same logic applies to major digital platforms, particularly Meta and social networks. In each case, users exchange a portion of their data for a service, often without precisely measuring the value of what they are transmitting.
Practical Steps to Limit Exposure
However, several habits can help limit data exposure:
- Obtain Consent: Before sending a photo of another person to an AI tool, it is best to obtain their consent, with particular vigilance for images of children.
- Review Permissions: Check the permissions granted to applications and, whenever possible, prefer access to specific photos rather than the entire gallery.
- Monitor Privacy Settings: Regularly verify settings related to the use of conversations and content, as options can vary from one service to another.
Conclusion: The Asymmetry of AI Usage
The real challenge is no longer just about knowing what AI can produce. As it becomes a daily tool, we must also ask ourselves what we are giving to achieve those results. The #Trend80 is just one example among many. A photo published for a few seconds of nostalgia can also be, for a platform, a new piece of data to process. It is this asymmetry between the apparent simplicity of usage and the complexity of what happens behind the screen that now deserves greater attention.
