Is Your Editing Software Using AI?

A Practical Guide for Photographers Entering International Salons

Why This Matters to You

If you use Lightroom, Photoshop, Topaz, DxO, Luminar, or just about any modern editing software, there is a good chance you are already using artificial intelligence in your workflow whether you realise it or not.

That is not necessarily a problem. Many AI-powered tools are perfectly fine to use in salon competitions. But some are not, and the line between the two is not always obvious from the software interface. Two buttons sitting right next to each other in your editing application might look and feel similar, yet one works within the rules and the other could lead to sanctions ranging from a three-year ban to permanent exclusion.

This article is a practical heads-up to help you think about what is happening behind those buttons, so you can make informed choices before you submit.

What FIAP and PSA Have Agreed

In 2025, FIAP and PSA jointly issued a statement that now must appear in the regulations of every salon held under their patronage. It draws a clear distinction:

Permitted: AI tools that work exclusively on the existing pixel data captured in your original photograph: enhancing, correcting, or transforming what is already there.

Prohibited: AI tools that introduce externally sourced content: visual elements, textures, objects, or scenes not originally present in your photograph.

In plain language: if the AI is helping you work with what your camera captured, that is fine. If the AI is bringing in something new from outside your image, that is not. The principle is straightforward. The practical application requires a bit more attention.

Assistive AI vs Generative AI

It helps to think of AI editing tools in two broad categories.

Assistive AI tools analyse and improve the data your camera sensor recorded without adding anything from outside the image. Think of them as very smart versions of tools photographers have always used, such as noise reduction, sharpening, and masking, just far more effective because the AI has learned patterns from training data. The key point is that these tools work within the boundaries of your original capture.

Generative AI tools use AI models , often trained on millions of external images, to synthesise entirely new visual content. When these tools operate, they generate pixels, textures, objects, or even entire scenes that were never present in your original photograph. The content might look seamless and natural, but it was not captured by your camera.

Type of AI	How It Works	FIAP/PSA
Assistive AI	Analyses and enhances the pixel data already captured by your sensor. No new visual content is introduced from outside the image.	Permitted
Generative AI	Uses AI models (often trained on millions of external images) to synthesise entirely new visual content: pixels, objects, textures, or scenes that were never in your original capture.	Prohibited

The Problem: We Cannot See Inside the Box

Here is something important that photographers need to understand: we do not always know exactly what is happening behind the button we click. The internal workings of most editing software are proprietary. Software companies do not publish the full details of their algorithms. A feature labelled as a simple “remove” or “enhance” tool might, under the hood, be using generative AI techniques that draw on externally trained models or it might not. The label on the button does not always tell the full story.

This means that listing specific tools as definitively “permitted” or “prohibited” is risky. Software updates can change what a tool does without changing what it is called. A feature that was purely assistive in one version could incorporate generative elements in the next. As photographers, we need to stay informed and exercise caution rather than rely on a fixed list.

Practical Clues (But Not Foolproof Tests)

While there is no guaranteed way to know exactly what an AI tool is doing internally, there are some practical observations that can help you stay alert.

Clue 1: The Internet Connection

Most generative AI processes are computationally intensive. Many of them are performed in the cloud, meaning your software sends data to remote servers for processing and receives the result back. If you disconnect from the internet and a particular editing feature stops working or fails to complete, there is a reasonable chance it relies on cloud-based generative AI.

However, this test is not 100% reliable in either direction. On one hand, some software is entirely cloud-based or web-based . The whole application runs on the company’s servers, not on your computer. If a feature stops working when you go offline, it might simply be because the software itself needs an internet connection to function at all, not because that specific feature uses generative AI. Many modern editing platforms operate this way, and being cloud-based does not automatically mean generative AI is involved.

On the other hand, there is the rapid rise of edge AI , which is AI processing that runs locally on your own computer’s hardware, particularly if you have a capable GPU. Some software already performs sophisticated AI generation entirely on your machine, without any internet connection. As GPU technology advances and AI models become more efficient, we can expect more generative AI features to run locally. The internet test will become less reliable over time.

Clue 2: Credit-Based Payment

Another signal worth paying attention to is how a feature is paid for. Generative AI typically requires significant computational resources, and software companies often pass that cost on to users through a credit-based payment model. You may receive a limited number of free credits to try a feature, but once those run out, you need to purchase more to keep using it. If a feature in your editing software works on a credit or token system, that is worth investigating further, as it may indicate that generative AI is involved behind the scenes.

But once again, caution is needed. Payment alone does not prove that a feature uses generative AI. A monthly subscription fee, for example, is simply a software licence. You are paying for access to the application, not for generative AI processing. The credit-based model is a useful clue, but it is not definitive proof.

The point of these clues is not to give you a simple pass/fail test. It is to encourage you to think critically about the tools you use and to ask questions when something does not feel clear.

Other Things to Be Aware Of

Different Modes Within the Same Tool

Some editing tools offer multiple operating modes. One works only with your image’s own pixel data, and another that invokes generative AI. These modes may sit side by side in the same panel, or the tool might default to a mode you did not intend. Before using any content-removal, fill, or expansion feature on a salon image, take a moment to check whether a generative AI option is enabled, and if so, switch it off.

Sky Replacement Depends on the Source

Replacing a sky using a built-in library or stock image introduces externally sourced content, which is prohibited. Compositing your own sky from a separate photograph you personally captured is a different matter, but the rules vary by salon division. In Nature and Photojournalism sections, almost any content modification beyond cropping and tonal adjustments is prohibited regardless of how it is done.

AI-Assisted Editing: Chatbot or Generative AI?

One of the newest developments in photo editing is the rise of AI assistants, chatbot-style interfaces built into your editing software that let you describe what you want in plain language. Adobe’s Photoshop AI Assistant, which entered public beta in March 2026, is a prominent example: you can type something like “make the sky more dramatic” and the assistant identifies the right tools and applies the changes for you.

The question for salon photographers is: is the AI assistant simply adjusting sliders and applying conventional edits on your behalf, or is it invoking generative AI behind the scenes? If the assistant is just automating the same adjustments you could make manually, such as moving an exposure slider, applying a curves adjustment, or selecting a region for you to edit, that is essentially a workflow convenience. But if the assistant is calling on generative AI tools to carry out your request, the result could be a prohibited image without you even realising it.

That said, if the AI assistant clearly tells you which actions and tools it is employing, and you can verify that those actions involve only ordinary adjustments rather than generative AI, then using the assistant should be fine. The key is transparency: you need to know what the assistant is actually doing, not just what it appears to be doing.

Where the assistant does not disclose its methods, or where you cannot be sure whether generative processes are involved, it is safer to make your edits manually using tools you understand. The conversational interface can make it feel like you are simply asking for help, but photographers entering FIAP salons need to be alert to what is happening behind the scenes.

In-Camera AI: A Growing Question

Modern smartphones and some mirrorless cameras apply AI processing before you ever open an editing application. Computational photography features that merge multiple frames, enhance details, or reconstruct data at the point of capture are becoming standard. Technically, in-camera AI should comply with the FIAP/PSA statement, because the processing works on the pixel data captured by the sensor without introducing externally sourced content.

But here is the deeper concern. We are already seeing hardware-level AI, dedicated AI chips built into cameras and smartphones that process image data as it is being captured. As edge AI becomes more advanced, there is a real possibility that even RAW files could be processed by AI before they are saved. Some devices may not retain a truly unedited, pure capture before AI takes place. If that becomes the norm, the assumption that a RAW file represents an unaltered record of what the sensor captured may no longer hold.

This does not mean in-camera AI is currently prohibited. By the letter of the FIAP/PSA statement, it appears not. But it is an area that the photography community will need to grapple with as the technology evolves, and as photographers we should be aware that this conversation is coming.

Six Simple Habits to Protect Yourself

1. Keep your RAW files. Always retain the original, unedited RAW file for every salon submission. This is your best evidence of authenticity.

2. Leave EXIF data intact. Do not strip metadata from salon entries. Complete EXIF supports your claim that an image originated from a real camera capture.

3. Enable your editing history log. Some software allows you to record every step of your editing workflow. Adobe Photoshop, for example, has a History Log feature that saves a detailed record of every action to the file’s metadata or a separate text file. If you enable this before you begin editing, you create a verifiable trail of what was done to your image.

4. Check your tool settings. Before using any content-removal or enhancement tool, check whether a generative AI mode is enabled. If you are unsure, try the feature with your internet disconnected and check whether it uses a credit-based payment model. These are not definitive tests, but they can raise useful flags.

5. Read the Conditions of Entry. Different salons and divisions may have additional restrictions. Nature sections are far more restrictive than Open sections regardless of AI.

6. When in doubt, do not use it. If you are unsure whether a tool crosses the line, simply do not use it on salon images. The penalties for getting it wrong far outweigh the convenience of any single edit.

A Note on Content Credentials and C2PA

One development worth watching is the emerging C2PA standard (Coalition for Content Provenance and Authenticity). Content Credentials, built on this standard, provide a cryptographically signed record of an image’s provenance, including which camera or software was used, what edits were made, and whether AI tools were involved at any stage. Think of it as a tamper-evident “nutrition label” for your photograph.

Adobe has built Content Credentials support into Photoshop and Lightroom, allowing users to apply provenance information at export time. In Photoshop, this currently supports JPG and PNG exports, while Lightroom currently supports JPEG export. On the capture side, the Leica M11-P integrates C2PA-based Content Credentials in-camera, and Sony’s PXW-Z300 can embed C2PA-compliant authenticity data at capture when its authenticity feature is enabled. Meanwhile, Google’s Pixel Camera app on the Pixel 10 lineup has achieved Assurance Level 2, currently the highest security rating defined by the C2PA Conformance Program.

For salon photographers, this is potentially very useful. A complete Content Credentials chain, from camera capture through editing to final export, provides strong evidence that your image originated from a real photograph and that no generative AI was involved. However, adoption is still growing, and the standard is not yet universally supported across all cameras and software. It is also worth noting that C2PA metadata can be stripped from files, so it is not a complete solution on its own.

Nonetheless, if your camera and software support Content Credentials, enabling them is a worthwhile habit. As the standard matures and more organisations begin to recognise it, having that provenance record attached to your salon entries could become an important advantage.

Concluding Remarks

AI technology is evolving rapidly, and so are the policies governing its use. PSA has already acknowledged that its AI rules are subject to change. The distinction between assistive and generative AI will continue to blur as software becomes more sophisticated, as edge AI makes local generative processing increasingly common, and as hardware-level AI in cameras challenges our assumptions about what constitutes an unaltered capture.

As photographers, we cannot afford to set and forget our understanding of these tools. Staying informed is not just about compliance. It is about understanding the medium we work in. One of the core purposes of the APS AI & Authorship Standards Subcommittee (AASS) is to educate our members about the evolving relationship between AI and photography. We continue to monitor developments across FIAP, PSA and other international bodies to keep you up to date.

If you have questions about AI and your editing workflow, please reach out to me or to the AASS. We are here to help.

Disclaimer: This article is published for educational purposes only. Neither the author nor the APS AI & Authorship Standards Subcommittee (AASS) claims to be the authoritative or final interpreter of FIAP’s AI rules. The purpose of this article is to raise photographer awareness of the impact of AI on photography, particularly when entering salons held under FIAP patronage. One of the core purposes of the AASS is to educate APS members in this regard. Photographers should always refer to the specific Conditions of Entry for any salon they enter and consult FIAP and PSA directly for definitive policy interpretation.