Android Authority has uncovered evidence suggesting that OpenAI may soon begin watermarking images generated through the ChatGPT app. This discovery comes from a deep dive into version 1.2025.196 (beta), where a new feature appears to let users opt to “Save without watermark” from the three-dot menu located in the top-right corner of the image generation interface. The inclusion of this option strongly implies that watermarking may soon become the default behavior when saving AI-generated visuals.
While the watermarking feature hasn’t been officially confirmed or rolled out yet, this development hints at a broader shift in how OpenAI handles content attribution and copyright transparency within ChatGPT. Watermarks are commonly used to help track the origin of AI-generated material and ensure that viewers or recipients know when an image was created by a generative model.
One key question that remains is whether all users will have the ability to bypass the watermark. As Android Authority points out, this “Save without watermark” option may ultimately be gated behind a paywall — possibly reserved for ChatGPT Plus subscribers or higher-tier users. If that’s the case, free-tier users would likely be limited to downloading watermarked images only.
It’s important to note that OpenAI has yet to confirm the presence or rollout of this feature. As with many functions unearthed in beta app versions, the watermarking capability could be modified or removed before reaching a stable release. For now, the discovery suggests that OpenAI is at least considering more visible safeguards or branding for its AI-generated imagery — a move that aligns with broader industry conversations around transparency, ethical content usage, and misinformation.
