OpenAI’s ChatGPT recently launched a new feature that allows users to transform their photos into Studio Ghibli-style images. This trend quickly went viral on social media. However, one user’s claim caught everyone’s attention—they stated that ChatGPT itself can revert Ghibli-stylized images back to their original form.
The Ghibli-style transformation feature was launched on March 26, 2025, and users immediately started experimenting with it. The trend began when Seattle-based software engineer Grant Slayton transformed a family photo into a Ghibli-style image and shared it online. His post gained over 50 million views, spreading the trend worldwide, including in India.
User claim Ghibli-Style Can Be Reverted to Original
A social media user shared a video claiming that ChatGPT can revert Ghibli-style images to their original form. The user demonstrated the process on their device, after which the video went viral, raising concerns about privacy.
Following the video’s spread, users began questioning whether OpenAI is storing user data for model training, even if they opt out. If reversing stylized images is truly possible, it could pose a major privacy risk, as personal data could potentially be recovered from the modified images.
Some users tested the claim themselves, with one commenting, “I checked it myself. It only gives a 50% match—the body shape remains, but the face is not exactly the same.”
Can Ghibli-Style Images Really Be Reverted?
Experts suggest that reverting a stylized image is not that simple. The Ghibli-style transformation is a generative AI process, where the original photo is analyzed and modified into a stylized version. During this process, important details such as skin texture, lighting, and realistic colors are altered or lost. AI also modifies fine facial details, making it difficult to fully recover the original image.
For AI to reverse the process, it would have to reconstruct lost details, which is a complex task. This would require a “de-stylization” process, where the AI removes the Ghibli-style effects and regenerates realistic details. However, OpenAI’s policy does not mention any such feature.
Some advanced AI tools like Stable Diffusion can perform de-stylization, but even their results are not entirely accurate.
