Abstract Blind fault mapping requires dense observations over large areas, and distributed acoustic sensing (DAS) provides a viable solution using dark fibers. Seismic attenuation is sensitive to fault‐related structure but relies on reliable amplitude measurements. In DAS data, amplitudes are distorted by non‐attenuation effects, including non‐uniform coupling and small‐scale heterogeneity, limiting their direct use for attenuation imaging. Here, we propose an attenuation imaging method based on ambient noise cross‐correlation functions, in which double virtual shot gathers are used to separate attenuation‐related and non‐attenuation‐related amplitude contributions. Using 6 hr of noise data, we obtain an attenuation coefficient profile along a 7.2‐km DAS array with sub‐kilometer resolution. The image reveals three strong‐attenuation zones, two of which are likely associated with fault‐related structures characterized by strong scattered waves, while the third is more likely associated with a low‐velocity structure. These results highlight the potential of DAS‐based ambient‐noise attenuation imaging for large‐scale, high‐resolution fault exploration.