Has Gemini's most viral moment come at a cost of users' privacy? Google's Nano Banana craze is reminiscent of Ghibli images that flooded everyone's timelines not long ago.

If you are on social media, it's hard to miss the AI-generated images from your friends, acquaintances, and others on common themes, such as realistic 3D avatars and shoots in a saree.

These images are generated using Google's Gemini chatbot, which was recently upgraded with a new AI image model that aims to give users more control over photo editing. The upgrade is likely aimed at keeping the AI on par with rival OpenAI's popular AI image tools, created via ChatGPT.

Under the hood is Gemini 2.5 Flash Image, also known as Nano Banana, which was rolled out late last month and has gone viral in the last couple of weeks.

Sweeping the world

The latest craze for Google's Nano Banana images is reminiscent of the Ghibli-style pictures that flooded everyone's timelines not long ago. Last week, Google disclosed that more than 500 million "Nano Banana" images have been created globally since the launch, with India emerging as one of the top users.

It stated that users have created a variety of images, including anime, manga, or mecha figures; retro and black-and-white portraits; and giant statues of themselves, among others.

"As usual, all images created or edited in the Gemini app include a visible watermark, as well as our invisible SynthID digital watermark, to clearly show they are AI-generated. Use the thumbs up and down buttons on your generated images to give us feedback, which we'll use to make ongoing improvements to our safety measures and overall experience," it said in a post.

As mentioned-above, ChatGPT had its own viral moment when it allowed users to create Ghibli-styled images. With millions of users flocking to the AI platform to create and share such images, OpenAI chief Sam Altman disclosed that the company added one million users in an hour at one point. To put things in perspective, ChatGPT onboarded one million users in five days of its official launch.

"the chatgpt launch 26 months ago was one of the craziest viral moments i'd ever seen, and we added one million users in five days. we added one million users in the last hour," he said in a post.

For Google too, the viral photo helped the company score more in terms of user base and popularity. The Gemini app raced to the top of global app stores', along with a 45% month-on-month surge in downloads in September, according to TechCrunch, citing data from app intelligence firm Appfigures. The report noted that the Gemini app saw 12.6 million downloads in September so far, remarkably high from the cumulative 8.7 million downloads in the month of August.

Privacy nightmare?

Even though these Ghibli and Nano Banana images are very cool and social media-friendly, giving away personal information as sensitive as photos could also be a privacy nightmare.

For instance, several experts, including those from law enforcement, have warned that such tools can be misused, such as by making people click on malicious links or download apps that look like the official ones.

"Gemini automatically saves your activity under something called Gemini Apps Activity. This means the photos, files, or even voice inputs you upload can be stored, used to train Google's AI, and sometimes reviewed by human evaluators. While these reviewers don't see your name, the data can still stay in Google's systems for up to three years longer than most people expect," said Aaush Raj, co-founder of Lurny Innovative Labs.

Raj also cautioned that sharing personal IDs, sensitive images, or confidential files, the risks go up from human reviewers possibly seeing them to AI drawing conclusions you didn't intend.

Google does not advise users to submit sensitive or personal information as it could be used for review and training Google AI.

Users are recommended to read the terms and conditions of using Google AI studio. Here's what it says:

"Google AI Studio is the fastest way to start exploring and building with Gemini, our next generation family of multimodal generative AI models. Test prompts, get an API key, and go build.

The Google APIs Terms of Service, Gemini API Additional Terms of Service, and the Google Privacy Policy apply. Prompts and responses may be reviewed and used to train Google AI, so don't submit sensitive or personal information. Learn more about data use. Gemini can make mistakes, so double-check it…"

Lasya Ippagunta, Manager-Artificial Intelligence at cybersecurity firm CloudSEK, explains that while platforms like Gemini incorporate safeguards such as visible and invisible watermarks to ensure transparency in AI-generated content, they often do not clarify whether uploaded images are stored or used for further training.

Moreover, free-tier services, in particular, may retain user data for model improvement, creating potential privacy risks. If such data is exposed or leaked, users' original photos could be made public without consent.

"It's essential for users to stay cautious—always review the platform's data policies, avoid uploading sensitive or personal images, and use these tools only when you're confident your privacy and security are being respected," Ippagunta explained.

Raghav Gupta, founder and CEO at Futuretense, further cautions users about giving away their photos to these AI platforms. He highlighted that such things could lead to even deepfake abuse.

"A face is more permanent than a password and cannot be changed if it gets misused. Beyond the glamour, the uploaded images can be stored, analyzed, or used for model training. Over time, this creates a digital trail that users cannot fully control. While it looks like harmless fun, it can open the door to risks like identity misuse, profiling, or even future deepfake abuse," he added.

Even though there are many privacy risks involved, users are still flocking to such platforms to create and subsequently share their photos. The back-to-back trend clearly shows that there is an appetite for these easy-to-use, AI-driven photo editing platforms. Internet firms may come up with more unique approaches that allow users to have fun with these tools without having to compromise or worry about their privacy.

As Gupta suggests, turning off account syncing or linking when possible reduces the data footprint. Most importantly, people should treat AI photo edits as they would any other online sharing.

"The same caution applied before posting personal images on social media should apply here as well. By being selective and privacy-aware, users can enjoy the nostalgia without unknowingly giving away their digital identity," Gupta said.