I belong to a photo club. The photo club has monthly salon competitions. Up ‘til the new release of Lightroom and Photoshop, the rules were very simple – essentially “do your own work” and “don’t add anything to an image that you don’t own”…

AI changed things. So, I created a proposal to add to the existing rules. Here’s the relevant piece

Images generated wholly, or in part, by user-initiated prompt Generative AI are prohibited, however, all available tools, including AI tools (i.e., generative fill), may be used to optimize one’s own captures provided no perceptible elements not captured by the photographer are introduced.

My , in the almost complete absence of information on how the tools work in context of a photograph was this:

My basis for believing generative fill without a prompt would generally work within the context of a photograph, and was not some monster coming to corrupt our images with random elements from some data library in the bowels of Firefly, was this:

When you run generative fill without a prompt, it displays tips. This is one of them. Seems to indicate that without a prompt generative fill WILL, if possible, use the surroundings – which I take to mean the current image – to fill.

Needless to say, the proposal caused a bit of disagreement. Mostly by one person who purports to have a heavy understanding of AI. His claim is that gen fill ALWAYS goes out and comes back with “stuff” from what Firefly has been trained on. And no matter HOW MUCH it matches what’s in your image, it actually has filled with stuff from this magical Adobe Library in Firefly. AND, that by the mere fact that you RAN the Generative Fill, YOU NOW OWN anything it returned.

My statement that I believed members of the club were ethical enough to stay within the guidelines and NOT to use a visible new element returned by unprompted gen fill was met with derision and the assertion that members encompassed a “wide range” of ethics and therefore COULD NOT be allowed access to technology they would surely misuse.

The person was ADAMANT that Generative fill IN ANY FORM must not be allowed to be used in ANY context for salons. Backed by a small bloc of like-minded people, the proposal was defeated. The proposal, however, is not dead and CAN be sent to the membership for a vote to override the board’s decision.

SO, long trip to the question: Does anyone in here have actual knowledge of whether or not generative fill, used without a prompt WILL or WILL NOT restrict itself if it can to the current image (the surroundings) and not pull all manner of non-allowed elements into your image? Or is it actually out to destroy photography by filling anything it touches with elements from images in some library from which the AI was trained? In which case the "tip" above about running gen fill is “misleading”?

Thoughts?