Django CVE-2026-53877

Exploitation requires that the target Django application (1) has django.contrib.gis installed and active, (2) constructs a GDALRaster object from a bytes argument where those bytes are influenced by attacker-controlled input, and (3) subsequently accesses the vsi_buffer property on the resulting object. The CVSS 4.0 AT:P metric confirms these specific attack requirements must be met. Applications that do not use the django.contrib.gis.gdal GIS module are entirely unaffected. Standard Django deployments without geospatial functionality carry zero exposure. The limiting factors for realistic exploitation are the requirement for attacker-reachable bytes input to GDALRaster and the subsequent vsi_buffer property access - both of which are application-specific code paths, not default Django framework behavior.

The CVSS 4.0 vector (AV:N/AC:L/AT:P/PR:N/UI:N/VC:L/VI:N/VA:L) scores 6.3 Medium. The AT:P (Attack Requirements: Present) metric is the critical moderating signal: exploitation requires the target application to actively use django.contrib.gis.gdal.GDALRaster with attacker-influenced bytes input, which is not a default Django deployment scenario. This meaningfully reduces the realistic attack surface compared to a base AV:N/PR:N read. The dual impact - VC:L (memory disclosure) and VA:L (potential segfault causing service disruption) - is concerning but bounded. There is no EPSS data available to quantify exploitation likelihood in the wild, no CISA KEV listing, and no public POC, all of which reduce urgency relative to the CVSS score alone. Organizations not using django.contrib.gis are unaffected entirely. Those running Django GIS applications that accept raster data from untrusted sources face the most realistic risk.

An attacker sends a crafted bytes payload representing a malformed raster object to a Django web endpoint that passes user-supplied data to GDALRaster(). When the application subsequently accesses vsi_buffer on the resulting object, the over-read either returns fragments of adjacent heap memory - potentially containing session tokens, keys, or other in-process data - or triggers a segmentation fault that crashes the worker process. No public exploit code is available at time of analysis.