Can we trust scientific images in the era of AI?

The proliferation of AI-generated scientific images is eroding public trust in science, according to a researcher studying visual science communication. While AI tools can aid scientists in creating illustrations, synthetic data, and educational materials, they also blur the lines between illustration, enhancement, and fabrication. This blurring has led to instances of AI-generated figures with biologically impossible structures appearing in scientific papers, resulting in retractions. For example, two papers were retracted in 2024 due to AI-generated figures, and in April 2026, the New England Journal of Medicine retracted a paper after discovering AI manipulation of a clinical image. These high-profile cases suggest a broader, less visible problem, with warnings that AI-generated visuals pose significant threats to fields reliant on visual evidence, such as materials science. Academic publishers are exploring AI-detection tools, but these systems are expected to perpetually lag behind the advancements in AI image generation technology.