People think Google is feeding their photos to Gemini, but the reality is worse
I recently came across an interesting discussion about the strange things happening in Google Photos: children being made up as oil paintings or hand drawings, etc. People suggested it was most likely an AI filter, not an AI-generated image, but that got me thinking about what Google is actually doing with our pictures. I'm not the only one who has been feeling weird about all the automatic AI manipulation happening in Google Photos. So, I looked into whether it was happening to me and what could be done to prevent something like it from happening in the future. What I found wasn't exactly expected.
People think Google is feeding their photos to AI But I couldn't find any evidence on my account It's clear that many people feel like Google is aggressively pushing Gemini to the point where it's forcing them to flee to other search engines. My experiences with it range from utterly useless to just plain bizarre. I get what they're trying to do with it, but nothing I've seen or experienced so far has convinced me that this is the future. With the aggressive Gemini push comes a whole host of other concerns, but privacy is the most significant one. When I read about this, I immediately checked my Google accounts to see if my data was being used for the training as well.
After reading the discussion I mentioned above and a dozen other articles about the exact same subject, I grew increasingly concerned that my photos were being used to train AI. I'd been using Google Photos for years prior to the big Gen AI rollout. Google has pictures of my cats (and my wife, too). I don't want them to be used to train Gemini, only to then be used as inspiration for so-called "AI Art" slop. But here's the interesting thing I found, personally, about the Google AI photos thing: I looked in my settings, and it turns out it never activated, nor does it even appear in Photos as a possibility on either of my phones (a Motorola Moto G and Samsung Galaxy Z Fold 5). So, I dug a little deeper. The feature discussed in the thread was Google Photos' "gallery-worthy" memories. You can turn it off in Settings > Preferences > Memories. From there, you can go further from there by turning off notifications and choosing Memory types. Here, you can shut off both Time-based and Themed memories. But at the bottom, there's another little option called Creations. Navigate there and shut everything off. That should get rid of any lingering "gallery-worthy" weirdness popping up with photos inside the Google Photos app. I couldn't find where it was switched on. There is a good chance that I shut it off without realizing it. I went through a period in 2025, where I began purging any accounts I wasn't using. Part of that included Gemini and old Google accounts. Although I don't quite recall doing it, I may have shut off everything AI-related in my Google account. Which is why I didn't notice any difference until I started experimenting with Gemini this year.
Google says it doesn't train AI on your photos Exploring the "Outside of Google Photos" loophole According to a Forbes report from December 2025, Google states it's not using user photo albums to train its AI. On December 23, 2025, Proton posted on X: When you know the only reason Google’s AI is the best at generating images because they’re scanning every Android user’s Google Photos albums but they won’t admit it & you can’t prove it. This set off a chain reaction of people wondering, does Google use our photos for AI training? Google says, "No, we don't; honest." They even posted it on a prominent answers page:
So, let's break that down. Google explicitly states they don't train outside of Google Photos. So that could potentially mean they'll absolutely use anything you move out of Google Photos to somewhere else in their ecosystem for AI training. I'm no lawyer, but...I'm definitely wary of that particular wording. It could also mean that they're using your photos to train the internal AI for Google Photos itself. If you move that photo through Gmail, or Drive, or Keep, it's subject to that particular app's rules governing use for AI training. If you connect photos to your smart-home display or some other third-party app, their warning is also telling you that since you made the data bridge, they're not responsible for what happens to the data that crosses over to it (in this case, your photos). Google does say they use AI to "personalize" your Google Photos. Last year, Google made a big push to change its privacy settings for Gemini on September 2, 2025. The privacy change allowed Google to use anything you've said, done, prompted, or uploaded to its Gemini to use for training purposes. They added a toggle called Keep Activity inside of Gemini (not to be confused with the Google Keep app). So keep that in mind when you save/sync something in Google, thinking it's private. Opting out is essential Take control of your images, consent, and privacy I am of the firm belief that users should have the option to opt in to something rather than needing to find a way to opt-out because something is switched on by default. While it seems like some features went that route (their new Personal Intelligence service is entirely opt-in). From the description in the original Reddit thread, it sounds like this might have been what they were experiencing. So, to nip it in the bud right now, you can go to gemini.google.com, click on Activity in the settings. Once there, click the button labeled On button and select Turn off and delete activity (2 steps). Then a warning will pop up. Select Got it, and you're done. Gemini, of course, weighed in on the subject:
It seems Google Photos' AI features aren't invading privacy the way you might think they are. What it's doing feels like it's something much, much worse. Just because my accounts weren't affected doesn't mean it isn't real Google seems to be adamant about not using our data to train its AI without consent, but reality and stories from across the internet tell a very different story. I wonder how many accounts were affected? And what new services are going to come around to further slurp up all our data for their training? Newspeak in the way they communicate might fool some people, but the best thing to do is be wary, find a way to opt-out, think critically, and always vet your sources.
Source: How-To Geek