The earliest known manipulated images originated in the 19th century, and Stalin famously had rivals edited out of photographs in the 20th century. Once digital tools became commonly available in the 1990s, the ease and extent at which photographs could be manipulated greatly expanded. This brings us to today’s age of AI proliferation, which has put the power of image generation and manipulation at our fingertips at a speed and scale never seen before. These new tools are transformative, but within the context of press and documentary photography, they need to be carefully regulated.
To explore this development, World Press Photo joined a working group that included representatives from Magnum Photos and the National Press Photographers Association, educational institutions, as well as independent professionals with interest and expertise in the topic. [Addition for clarification: These organizations and individuals were not consulted at any time about World Press Photo Contest rules.]
Together we developed a set of
clear ethical standards to help guide photojournalists, documentary photographers, photo editors and media organizations in response to the emergence of AI image-generation tools. These standards form the basis of our own response to new AI tools in the annual
World Press Photo Contest.
When is a photograph not a photograph?
One of the realizations we came to is that generative AI tools cannot create photographs. They can do amazing things, and can create photorealistic or "photo-like" images that are easily mistaken for actual photography.
A photograph, by contrast, captures light on a sensor or film. It is a record of a physical moment.
This (simplified) definition of photography does not resolve all ethical dilemmas, of course. Photographers choose what to point their cameras at, how to frame their subjects, which photos to share or publish. Events can be staged or faked. Photos are edited (using AI tools or other techniques). Captions and other context surrounding the image can either add to an accurate understanding of the image or mislead the viewer into thinking it’s something else entirely.
This means the source of the image is also important. The photographer, photo editor and publisher need to be trustworthy. We take our role in this extremely seriously. Whether online or at an exhibition, people should be able to trust every World Press Photo Contest winning story.
What changes and does not change in our contest
Singles, Stories and Long-Term Projects
There are no real changes to these categories, but we have adapted our rules to make it more clear what is and is not acceptable.
All photographs entered into the Singles, Stories and Long-Term Projects categories must be made by a lens-based camera. No synthetic or artificially generated images are allowed, and no use of artificially generative fill is allowed in post-production. Any use of these tools will automatically disqualify the entry from the contest.
However, the use of AI-powered enhancement tools is possible within the contest rules, as long as these tools do not lead to significant changes to the image as a whole, introduce new information to the image, nor remove information from the image that was captured by the camera.
Some examples of tools where limited usage may be allowed are Denoise, automatic adjustments (e.g on levels, colors, contrast) and object selection. These are permitted up to a certain extent, which is to be determined by the contest organization and global jury. Tools that do immediately breach the contest rules are all AI powered enlarging tools such as Adobe super resolution and Topaz Photo AI. These tools are based on generative AI models that introduce new information to enlarge and sharpen images.
Open Format
The purpose of the Open Format category is to encourage entries that feature innovative techniques, non-traditional modes of presentation, and new approaches to storytelling. We think it is important to have space for this in our contest because it creates the opportunity to showcase positive examples and spark discussion in a rapidly changing industry.
Many of the winners in the short history of this category have overlaid graphical elements and text on their images. This is just one of the techniques permitted under our rules. It would not be logical to put limitations on what tools can be used to create these non-photographic elements.
Therefore, we will permit the use of generative AI in the Open Format category, as long as entries incorporate lens-based still photography as the source and central part of the work. Entries that solely contain artificially generated images will be disqualified from the contest.
While entering work, entrants will be asked to specify whether generative AI has been applied and to what extent. All Open Format entries will then be subject to a thorough entry check process by the contest organization to determine whether they meet the above criteria.
We recognize that these rules themselves are being tested, and we evaluate and improve our rules every contest season.
How do we prevent cheating?
We believe we have the most thorough and careful process of any photo competition. At the same time, we have to acknowledge the new challenges. Nonetheless, we still have confidence in our process.
As before, we will use a multi-layered defense against fakes.
- Only professional news and documentary photographers may enter.
- All entries must have accurate captions and descriptions that answer the five basic questions of good journalism (who, what, where, when, why).
- Judging is done by professionals who consider the story as well as the image.
- At a later stage in the judging, original files are requested and examined by two independent digital analysts.
- Once winners are selected, an external research team will check the context in which each image or story was produced, distributed and/or published.
Why is all of this important?
Photography needs to remain a credible witness to world events. Extensive digital manipulation of photographs over the last three decades, along with the advent of photorealistic imagery generated by artificial intelligence systems, put this ability at risk. This is not a challenge World Press Photo can overcome on its own. But we are determined to do our part so that the millions who visit our exhibitions around the world, and the millions more who see these winning photos online, can have full faith and confidence they are getting an accurate and representative view of our shared world.
