AI skepticism threatens trust in real news photos, journalist warns

A photojournalist's experience covering a school shooting in Minneapolis revealed a disturbing trend: her authentic images were met with skepticism and accusations of being AI-generated. Nora Savosnick, who has spent a decade covering conflict and politics, captured a poignant moment of two friends embracing after the attack. Shortly after publication, she received emails questioning the authenticity of the photos, with one sender stating, "I think it's AI. ChatGPT thinks so too."

“

The most dangerous thing about AI is not that fake images look real. It is that we dismiss real news images as fake.

This incident underscores a growing concern: the danger is not just that fake images look real, but that real news images are dismissed as fake. Savosnick possessed metadata, before-and-after shots, video, and audio evidence to prove the images' authenticity. However, she notes that metadata is fragile, easily altered or removed, and often doesn't travel with the image online. The core problem, she explains, is not a lack of proof, but that the proof doesn't accompany the image where it is consumed.

The proliferation of AI-generated visuals, propaganda, and satire on social media makes it increasingly difficult to discern what is real, manipulated, or AI-generated. Savosnick observes that people tend to believe images that align with their existing beliefs and dismiss those that contradict them. This dynamic creates a "liar's dividend," where the mere possibility of AI generation allows truth to be sidelined. The standard default of suspicion erodes trust in genuine reporting.

“

When everything can be called artificial intelligence (AI), truth can be pushed aside.

To combat this, Savosnick argues that a simple caption is insufficient. In the non-linear online environment, images, text, and evidence must be integrated. She proposes that news images need a form of interactive, secure "table of contents" embedded within the image itself. This would clearly indicate who took the photo, who published it, when and where it was taken, and what edits, if any, were made. This transparency, she suggests, is crucial for rebuilding credibility in an era of advanced AI manipulation.

“

Can you provide some form of confirmation that the image is real?