Large Language Model-Generated Content Policy

The Geneseo Current occasionally uses large language models(LLM) to assist in creating graphics and illustrations.

We recognize that generated imagery is becoming increasingly common and believe readers should be able to distinguish between an illustration and an image documenting an actual event.

Our Standard

Generated graphics are used as visual illustrations. They are not intended to represent documentary photography or suggest that a depicted scene actually occurred.

We will not use LLM’s to create an image of a real local event and present that image as authentic news photography.

When a generated image could reasonably be mistaken for an actual photograph or documentation of an event, we will identify it as LLM generated or an LLM generated illustration.

Graphics that are clearly illustrative in nature may not carry an individual disclosure when there is little reasonable possibility that they could be mistaken for documentary photography.

News Photography

Photographs presented as coverage of actual local events are expected to represent those events.

We will not use an LLM to add, remove, or substantially alter people, objects, actions, or circumstances in news photography in a way that changes the factual meaning of the image.

Accuracy Still Matters

LLM generated graphics accompanying a news story should not introduce factual claims that are unsupported by the reporting.

When a graphic contains factual information—such as statistics, dates, locations, quotations, or other reporting—the information is subject to the same accuracy standards as the accompanying story.

Transparency

Our goal is straightforward: readers should not be misled about whether an image represents something that actually happened.

When that distinction may not be obvious, we will disclose the use of large language models(LLM’s).

Questions regarding the factual accuracy of our reporting or graphics are always welcome.

The Geneseo Current uses large language model tools as part of our modern newsroom and business operations. We believe these tools can improve efficiency, strengthen research, organize information, and help us better serve our readers.

We also recognize that LLM’s can and do make mistakes, misunderstand context, generate inaccurate information, and fail to distinguish reliable information from unreliable information.

For that reason, LLM’s does not replace human editorial judgment.

Large Language Model Generated Images and Visuals

The Current may use large language model(LLM) -assisted or LLM-generated illustrations and graphics.

LLM-generated imagery should not be presented in a way that reasonably leads readers to believe it is documentary photography of an actual event.

When a LLM-generated image could reasonably be mistaken for a real photograph, we will identify it as an AI-generated image or illustration.

We will not use a LLM to materially alter a journalistic photograph in a way that changes the factual meaning of what occurred.

Routine photographic adjustments such as cropping, resizing, exposure correction, or other basic editing are not considered LLM-generated journalism under this policy.