Why You Should Stop Doing the ChatGPT 80s Trend Right Now

If you've done the viral 80s photo trend on ChatGPT, here's the direct answer: your selfie is very likely being used to help train future versions of the model right now, today, unless you went into a settings menu you probably never noticed — because on a personal ChatGPT account, that setting is turned on by default. The counterintuitive part almost nobody mentions: even if you already found that setting and switched it off, giving your 80s photo a simple thumbs-up because you loved how it turned out can quietly pull that exact conversation back into the training pile anyway.

Jake had already done the trend twice by the time a customer asked him about it. He runs a small phone repair and resale shop, and between fixing cracked screens he'd uploaded a photo of himself to ChatGPT and gotten back a warm, grainy portrait that looked like a 1985 mall studio sitting. He posted it to his shop's Instagram because, in his words, "it made me look like I had way better hair in 1985." Then a regular customer picking up a repaired phone asked him, half-joking, "Did you just hand your face over to a computer company?"

Jake didn't have a good answer. So he asked Ethan, who helps him with the shop's IT and has strong opinions about most of it.

What the 80s Photo Trend Actually Is, and How It Spread

The trend picked up speed on Instagram in early September 2026, spreading through Reels, Stories, and "Add Yours" stickers. The idea is simple: you open ChatGPT, attach a clear, well-lit selfie, and type a prompt asking it to reimagine you in an authentic 1980s style — big hair, period clothing, warm studio lighting, visible film grain — while keeping your face recognizably yours. People have run the same idea on parents, kids, pets, and even public figures, and the results get shared widely because they look genuinely convincing.

It's not the first trend like this. A very similar wave happened with Studio Ghibli-style portraits, and before that, AI caricature generators went through the same cycle: a fun idea, a flood of uploads, then a second wave of articles asking whether anyone stopped to read the fine print. The pattern repeats because the underlying mechanics never really change — you're handing a company a clear image of your face, on purpose, at scale, along with everyone else doing the same trend that week.

Does OpenAI Actually Use Your Photo to Train ChatGPT?

Yes, by default, on a personal account. OpenAI's own help documentation states plainly that when you use ChatGPT as an individual, your content — including photos you attach — may be used to train its models unless you opt out. The setting that controls this is called "Improve the model for everyone," and it lives under Settings → Data Controls. On Free, Plus, and Pro personal accounts, data sharing is enabled the moment you create an account. Nobody asks you to confirm it during sign-up; it's simply the starting position.

Business accounts work the opposite way. ChatGPT Business, ChatGPT Enterprise, ChatGPT Edu, and the API do not train on your inputs or outputs by default. If your workplace pays for one of those tiers, your 80s photo experiment at your desk is treated very differently than the same experiment on a personal Plus subscription at home. That split — consumer accounts opted in, business accounts opted out — is the single most important fact in this entire post, and it's easy to miss because the app looks identical either way.

The court order that briefly changed everything

For a period, OpenAI was under a legal preservation order tied to litigation with The New York Times, which required the company to hold onto user data — including conversations users had deleted — rather than remove it within the normal window. OpenAI has since said it returned to its standard retention practice: deleted ChatGPT conversations and Temporary Chats are automatically removed from its systems within 30 days, and API data is deleted on the same 30-day schedule. The company noted it still has to securely retain a narrow slice of historical data from a specific 2025 window under that litigation, but that new conversations going forward follow the ordinary rules again.

The Data Hiding Inside the Photo File Itself

Here's a piece most 80s-trend articles skip entirely: the picture you upload usually carries more than pixels. Modern phone cameras embed a block of hidden information called EXIF data directly into every photo file — think of it as a shipping label stapled to the inside of the box, invisible unless someone opens it up. EXIF data can include the exact date and time the photo was taken, the camera or phone model, and, if location services were on, the GPS coordinates of where you were standing.

Most social platforms strip this out automatically before displaying your post — Instagram and Facebook typically remove it on upload. But when you attach a photo directly into a chat window to get your 80s portrait made, you're sending the original file, EXIF and all, straight to the company's servers before anyone strips anything. Whether that metadata gets used, stored, or discarded as part of processing isn't something you can verify from the outside; the safest assumption is that whatever was embedded in the file traveled along with it.

Why a Face Isn't Like a Password

Ethan puts it to Jake this way: "If someone steals your password, you change your password. Nobody can hand you a new face." That's the entire argument behind why privacy researchers keep flagging AI photo trends specifically, rather than treating them like any other online upload. A password, a credit card number, even a Social Security number — all of these can, in theory, be replaced. Facial geometry — the distance between your eyes, the shape of your jaw, the proportions that make your face recognizably yours — cannot.

That permanence is exactly why laws exist that treat facial data differently from ordinary information. Illinois passed the Biometric Information Privacy Act back in 2008, after a biometric payment company called Pay By Touch collapsed and people realized nobody knew what would happen to the fingerprints it had collected. BIPA requires a company to get written consent before collecting someone's facial geometry, retina scan, fingerprint, or voiceprint, and — unusually for a U.S. privacy law — it lets ordinary people sue directly if a company breaks that rule.

ChatGPT vs. Random "80s Photo" Apps and Websites

Every time a photo trend goes viral, a second wave follows it: copycat apps and single-purpose websites promising the same effect, sometimes faster or with no login required. Jake had a customer come into the shop asking whether he should try one of these instead of ChatGPT, since it didn't ask him to make an account at all.

Ethan's answer was blunt: "No account and no login isn't a privacy feature — it just means there's no record of who agreed to what." A named, publicly traded company with a published privacy policy, a help center, and a legal team answerable to regulators is not the same risk profile as an app that appeared in an app store two weeks ago riding a trend, with no clear indication of who built it, where the servers are, or how long anything is kept. This isn't a claim that every small app is malicious — most probably aren't. It's that you have no way to verify anything about the ones you can't research, while OpenAI's data practices are at least written down somewhere you can go read them yourself, which is exactly what this post has been doing.

How to Turn Off Training and Use Temporary Chat

These steps are the same whether you're on a phone, tablet, or computer — the settings menu looks the same across ChatGPT's apps and its website.

1. Open ChatGPT and sign in to your personal account.
2. Tap or click your profile icon, usually in the top corner of the screen, and choose Settings.
3. Select Data Controls.
4. Find the toggle labeled "Improve the model for everyone" and switch it off.
5. Close and reopen Settings a moment later to confirm the toggle held — this applies to every device signed in to the same account.

This change is forward-looking only. It stops new conversations from being used to train future models; it does nothing to conversations that already happened before you flipped the switch. If you want a conversation to never be used for training or even show up in your chat history in the first place, use Temporary Chat instead — it's a separate mode you turn on for a single conversation, and OpenAI states those chats are automatically deleted from its systems within 30 days and are never used for training, even if your main toggle is still on.

How to Delete or Export What You Already Uploaded

If you've already done the trend and want to clean up after yourself, here's the honest version of what deletion does and doesn't accomplish.

1. Open ChatGPT, go to your chat history, and find the conversation with your 80s photo.
2. Delete that specific conversation rather than archiving it — archiving hides it from your sidebar but keeps it in your account under the same retention rules.
3. If you'd rather see everything you've shared before deciding what to remove, go to Settings → Data Controls and use the export option to download your full ChatGPT history first.
4. If your account has ChatGPT's Memory feature turned on, check Settings for any saved memories connected to that conversation — deleting the chat itself does not automatically clear a memory that was extracted from it.
5. If you want everything gone, including your account itself, use the account deletion option in Settings, understanding it follows the same roughly 30-day removal window as an individual deleted chat.

A Pre-Upload Checklist for the Next Viral Photo Trend

Because there will absolutely be another one of these trends in a few months, whatever the decade or theme happens to be. Ethan's rule of thumb for Jake's shop customers is short: treat every new photo trend the way you'd treat handing someone your driver's license — quick to do, hard to take back, so it's worth a ten-second gut check first.

Photos of Other People, Kids, and Coworkers

This is the part of the trend that gets the least attention and probably deserves the most. When you upload your own selfie, you're making a decision about your own face. When you upload a group photo from a birthday party, a family photo with your kids, or a picture from a work event with coworkers in the frame, you're making that same decision on behalf of every other person in it — and none of them agreed to anything.

The general advice from privacy researchers and regulators tracking these trends has been consistent across every wave of them, from the AI yearbook trend a few years back through the Ghibli-style wave to this one: avoid using photos of minors in these trends at all, and think twice before uploading any photo that includes someone who hasn't been asked. A cute group photo you post for laughs is, from a data standpoint, several people's facial data submitted on one click, by one person who only had the authority to make that call for themselves.

When the Fix Doesn't Work, and What You Genuinely Can't Undo

Let's be straight about the limits here, because pretending every risk has a clean fix is worse than admitting a few don't. If your photo has already been shared publicly on Instagram, X, or Facebook, deleting the original ChatGPT conversation does nothing to the copies that have already been downloaded, screenshotted, or reshared by other people — that ship has sailed the moment it went public. If a completed model training run already included your image, there's no retroactive button that removes it. And if you don't remember which account, which email, or which device you originally used, you can't request deletion of data tied to an account you can no longer access or verify.

None of that means the trend is uniquely dangerous compared to everything else you already do online — most people have already shared far more sensitive information in far less careful ways. It means this specific trend is a good moment to actually go check a setting you've probably never opened, because now you have a concrete reason to.

Frequently Asked Questions

Is the 80s AI photo trend safe to try at all?

It's not inherently dangerous the way a scam link is, but it does mean handing a clear photo of your face to a company under whatever data settings your account currently has. It's a manageable risk if you check your settings first, not a reason to panic if you've already done it.

Does OpenAI keep my photo after I close the chat window?

Yes, unless you used Temporary Chat. A regular conversation, including any photo attached to it, stays in your account history until you delete it, and then it's scheduled for removal from OpenAI's systems within roughly 30 days.

Does turning off "Improve the model for everyone" delete my past photos?

No. It only stops new conversations from being used for training going forward. Anything sent before you switched it off may already have been used, and the toggle can't reach backward.

What happens if I give a thumbs up or thumbs down on my 80s photo?

OpenAI's own documentation states that submitting feedback can cause the entire conversation tied to that feedback to be used for training, even if your general training setting is turned off. Reacting to a specific result overrides the toggle for that one conversation.

Is Temporary Chat actually private?

It's more private than a regular chat: it won't show up in your history, won't be used for training, and OpenAI says it's automatically deleted within about 30 days. It isn't a promise of total secrecy from OpenAI itself — the company can still see it while it exists, the same as any conversation.

Does ChatGPT Business or Enterprise use my photos for training?

No, not by default. OpenAI states that Business, Enterprise, Edu, and API usage is not used to train models unless a workspace administrator specifically changes that setting.

What is EXIF metadata, and does ChatGPT keep it?

EXIF data is hidden information embedded in a photo file, like the date, camera model, and sometimes GPS location. Whether it's stripped or retained isn't publicly documented in detail, so the safe assumption is that whatever was in the original file traveled with it to the server.

Can someone find my home from an 80s AI photo I post?

Only if location data was embedded in the file and never stripped, and only if that data somehow became visible. Removing GPS data from the photo before uploading it closes that particular risk.

Are random "80s photo" apps and websites safer or riskier than ChatGPT?

Generally riskier, because you usually can't verify who built them, where the data goes, or what their retention policy is. A known company with a published, readable privacy policy at least gives you something to check.

Can my face be used to create a deepfake from this trend?

A clear, well-lit photo is exactly the kind of input that makes convincing manipulated images or videos easier to produce, which is one reason security researchers keep flagging high-resolution facial uploads specifically, regardless of which app or trend prompted them.

What is biometric data, and why does it matter here?

Biometric data covers unique physical measurements like facial geometry, fingerprints, or voiceprints. It matters because, unlike a password, it can't be changed if it's ever misused — you're stuck with the same face for life.

Does the 80s trend violate any privacy law?

Not automatically. Illinois's Biometric Information Privacy Act, for example, specifically excludes plain photographs from its definition of biometric identifiers — it targets extracted facial-geometry templates instead. It's also a state law, so it only applies in certain circumstances tied to Illinois, not nationwide.

Should I let my kids do the 80s photo trend?

Privacy researchers and watchdogs who've tracked similar trends consistently advise against uploading photos of minors to these tools at all — it's one of the more clear-cut pieces of advice in this whole topic.

What should I do if I already uploaded a photo and regret it?

Delete the specific conversation, turn off training going forward, and if you're on Memory, check for and clear any saved memory tied to it. Understand that deletion is forward-looking and won't undo anything already used in a completed training run.

Does deleting my ChatGPT account remove everything immediately?

No. Account deletion follows the same roughly 30-day removal window as deleting a single chat, and a narrow set of data may be retained longer for legal or security reasons.

Is there a safer way to get the 80s look without the privacy trade-off?

Using a Temporary Chat, stripping location data from the photo first, and choosing a well-known platform over an unfamiliar app gets you most of the fun with meaningfully less exposure — it doesn't eliminate the trade-off, but it shrinks it considerably.

Revision note. Written September 2026, covering ChatGPT's personal (Free, Plus, Pro) and business-tier (Business, Enterprise, Edu, API) data controls as they stood at time of writing, along with general biometric-privacy context that isn't tied to any single app version..But, Whatever trend comes next, hopefully this saved you the ten seconds it takes to check a setting before you send your face to anyone.Follow security practises seriously, period!