OpenAI seems to be getting ready to introduce watermarks on AI-generated photos in ChatGPT, based mostly on findings from an APK teardown of the app’s newest beta model (v1.2025.196). The characteristic, presently in growth, could apply watermarks to all generated photos, significantly for customers on the free tier. Nonetheless, indications recommend that paid customers may very well be given the choice to save lots of photos with out the watermark.
The APK teardown — a course of the place in-development app code is examined for unreleased options — revealed a string labeled “image-gen-watermark-for-free,” which means that watermarking could also be restricted to free accounts. This aligns with an earlier leak from April 2025, when a watermark characteristic labeled “ImageGen” was first noticed by a consumer on X (previously Twitter).
Screenshots from the activated characteristic shared by AndroidAuthority present the usual picture interplay choices — Edit, Choose, Save, and Share — together with a new choice within the three-dot menu: “Save with out watermark.” Whereas this characteristic was momentarily enabled even for a free-tier account throughout testing, it’s possible that entry can be restricted to subscribers as soon as the characteristic is formally launched.
Picture by AndroidAuthority
There’s nonetheless some uncertainty about how the watermarking system can be carried out. One chance is that all customers will see watermarked photos by default, with solely paid customers capable of take away the watermark. One other state of affairs might contain permitting all customers to save lots of unwatermarked photos, however by means of a extra sophisticated course of that daunts widespread informal use. This technique might assist OpenAI steadiness transparency and branding with consumer flexibility.
It’s essential to notice that this watermarking characteristic has not been formally introduced by OpenAI, and its inclusion within the ultimate launch isn’t assured. APK teardowns provide insights into options that builders are experimenting with, however they don’t verify ultimate implementation.
OpenAI continues to experiment with a variety of latest options for ChatGPT, together with Examine Collectively modes, Picture Types, and subscription modifications like a potential annual plan. The watermarking characteristic provides to this listing of in-progress enhancements, doubtlessly signaling a broader effort to handle the traceability and use of AI-generated content material.
Till OpenAI makes an official assertion, these findings stay speculative. Additional updates are anticipated as growth progresses.
Filed in . Learn extra about AI (Synthetic Intelligence), ChatGPT and OpenAI.
