80s trend: what are the risks of sending photos to AI?

If you've opened Instagram, TikTok, or WhatsApp in the last few weeks, you've surely stumbled across someone who decided to go back four decades in time. Big hair. Shoulder pads. Denim jacket. That yearbook-style photo, grainy like film and with slightly faded colors, as if it came straight out of a dusty drawer from the 80s.

The trend spread across social media too quickly to be ignored. People like Sabrina Sato, Virginia Fonseca, Fernanda Paes Leme, and Ana Maria Braga joined in, the hashtag #trendanos80 already has thousands of posts, and Google Trends itself registered the term among the trending searches in Brazil. Just send a selfie to ChatGPT or Gemini with the right prompt, and voila: you're back in the 80s in seconds.

This is the kind of phenomenon that helps explain why certain AI trends are growing so fast. The experience is simple, the result is beautiful to look at, and there's a huge social push: everyone, literally everyone, is doing it.

But there's one question that almost never appears next to the "generate" button.

What happens to your photo after you press send?

We showed our faces before thinking about what we were showing

There's nothing wrong with participating in the fun. Using an AI tool to reimagine a photo doesn't, in itself, mean giving up your privacy.

The point lies elsewhere: we think a lot about the final result— the hair, the clothes, the perfect filter—and almost nothing about the journey the image takes to get there. Is it stored? For how long? Can it be used to train other models? Is it shared with third parties? Can it be deleted later?

These are far less fun questions than choosing between a denim jacket and a colorful blazer. But these are precisely the questions that should factor into the decision, especially since experts have already begun to comment on this specific trend, and the warning isn't generic.

Your face isn't a password that can be changed

According to the specialized lawyer interviewed by SBT News about the trend, the risk of a leak may be low, but it exists, and the problem is not limited to AI tools: social networks and other digital services also concentrate a large volume of personal data.

Eliney Sabino, coordinator of technology courses at UnifacenS and AI specialist, reinforces a simple yet troubling point: those participating in the trend aren't just handing over "a photo," they're handing over data that can be cross-referenced with their name, social media profile, workplace, and routine to create a much more complete portrait of who they are.

And here lies the difference between a face and a password: a compromised password, you change it. A stolen card, you cancel it. Your face, you don't. There's no such thing as "generating a new version" of your physical identity if it's misused.

This doesn't mean that every selfie sent to an AI will become a biometric identifier used against you. It means that, before handing over potentially sensitive data, it's worth understanding the rules of the game for that specific service.

The difference seems small. In practice, it's enormous.

The number nobody notices: almost 1 in 3 people don't even read the terms

A recent finding, gathered from a study on digital behavior in Latin America, helps explain why trends like this are so concerning to security experts: approximately 27% to 30% of users in the region admit to not reviewing the permissions an app requests before installing or using it, either because the text is too long or because they simply don't pay attention.

In other words: most of us click "accept" on autopilot. And when a trend is booming, that autopilot kicks in even more. You see a friend doing it, open the tool, upload the photo, choose the style, publish. In less than a minute, everything is done. The invisible part— data retention, sharing, use for training—is left for later. If it's ever done at all.

What if your photo contains more information than you realized?

There is also a much more trivial risk, and therefore easy to forget: the photo you send may carry much more than just your face.

A name tag hanging around his neck. A computer screen in the background, with a spreadsheet open. A document on the table. The inside of his house. A car license plate. A child who did not (and could not) give consent to appear there.

A seemingly innocent selfie can reveal far more context than you intended to share, and experts interviewed by various media outlets all agree: before sending it, it's worth evaluating what the image shows around your face, not just your face itself.

From a joke on social media to the raw material for a scam

Here's where the layer that least appears in headlines about the trend comes in: the more clear photos of faces circulate publicly, the more raw material there is for deepfakes and social engineering scams, including corporate ones.

Analyses aimed at executive audiences already point to this issue: the trend represents a security risk that directors and IT teams should be aware of, because clear, high-quality photos—exactly the type the trend demands to function well—are also the type of material that fuels attempts to impersonate others.

A photo alone doesn't create a sophisticated scam. But, combined with name, position, company, voice, and routine (information already circulating), it can become a piece of a puzzle far more convincing than anyone would like to admit.

Read also: AI already believing: why it's so difficult to identify AI content

So, should we stop using these tools?

No. It would be an exaggeration to treat any image generation app as a threat to be avoided at all costs.

The point is to gain a little more awareness before clicking "send." In practice, this means:

• Take a quick look at the tool's privacy policy before uploading the photo;
• Check if it's possible to refuse the use of the image in model training, or request its deletion afterward;
• Avoid photos with documents, badges, work screens, or other people who did not ask to be there;
• For companies: turn this into a clear policy, specifying what kind of image and information employees can (or cannot) upload to external AI tools, because there's no point in investing heavily in system security and leaving this side door open.

You may also be interested in: How to prevent data leaks in AI: security strategies, LGPD (Brazilian General Data Protection Law), and governance.

The trend may pass. The image remains yours

This is perhaps the most interesting part of the whole story. In a few weeks, no one will remember the AI-generated denim jacket anymore; another aesthetic will be dominating the feed.

But the photos submitted to participate in the game don't disappear at the same rate as the trend.

The idea here is not to spoil the fun or demonize technology. It's to remember that, behind the "generate" button, there's data that remains yours long after the trend has faded.

Enjoying the nostalgia of the 80s? Always. But perhaps the most current way to protect in the age of AI is this simple: before any filter, ask yourself "what exactly am I giving in return?"

Written by Carlos Alcoba