A photograph of Earth taken during NASA's Artemis II mission in April 2026 went viral, evoking the iconic Apollo 8 'Earthrise' — but in an era when anyone can fabricate a nearly identical image with an AI text prompt, trust in scientific imagery is fracturing.
AI tools are already reshaping how scientific visuals are created: researchers use them to generate illustrations, synthetic data, edit lab photos, and craft outreach materials. The speed and realism of AI-generated images blur the line between authentic documentation and artificial fabrication.
This is not just a misinformation problem, according to experts studying visual science communication. It signals a deeper crisis of trust in science itself. The traditional markers scientists relied on — provenance, peer review, institutional credibility — are losing their power to convince a skeptical public.
When a single text prompt can produce a visually convincing Earthrise, people lose the ability to distinguish real exploration from synthetic spectacle. The implications extend beyond public confusion: they affect how scientific findings are communicated, funded, and believed.
The challenge ahead is not only technical — developing detection tools — but also cultural: rebuilding the visual credibility of science in an age where seeing is no longer believing.