Representation of retractions in scientific media

Retractions serve as a key mechanism for correcting scientific knowledge. Despite their importance, retractions can be associated with reputational concerns, and retracting a scientific publication is often perceived as an individual failure, rather than a routine procedure of scientific work. Discourse about retractions can play an important role in shaping the understanding of retractions. It may influence how researchers and the public more broadly think about the retractions process and how scientists correct mistakes.We report findings of a longitudinal automated content analysis examining the representation of retractions in the informal content (n=1016), e.g., news, editorials, opinion pieces, of two prominent scientific journals. We compare two text classification techniques for text analysis and demonstrate the superiority of text classification using GPT4.1 with API request (accuracy f1=0.86) compared to classification using the dictionary method (accuracy f1=0.67).Our analysis shows that discourse about retractions is heavily centered on discussions of unethical behaviors. We observe a media discourse dynamic where high-profile cases attract significant attention, overshadowing retractions made for errors and honest mistakes. While controversial narratives resonate with audiences, they may cultivate a skewed perception of retractions primarily as a symbol of scientific misconduct. This trend may contribute to the stigmatization of retractions and discourage researchers from retracting flawed work due to fear of reputational damage, ultimately undermining transparency and integrity in scientific research.