Loading vulnerability…
CVE-2026-15307 is a high-severity vulnerability in GeoDjango affecting Django 5.2 before 5.2.17 and 6.0 before 6.0.8. During spatial lookups against a GeometryField or RasterField, right-hand-side values are passed to the django.contrib.gis.gdal.GDALRaster constructor and are optimistically interpreted as raster input. Untrusted values can therefore reach GDALRaster through application-controlled lookup parameters, including spatial-field filters exposed through the Django admin changelist to a staff user with view permission. When the supplied value is a dict, or a string containing its JSON representation, GDAL opens it in write mode despite the constructor's write=False default, enabling file creation with attacker-controlled name and contents through a file-backed raster driver. Other attacker-controlled strings are treated as datasources, which can trigger outbound requests through GDAL virtual filesystem handlers. If the resulting file write lands in a location later imported or executed by the application, the flaw can be escalated to remote code execution.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
No public exploits tracked yet. Mallory keeps watching.
No public exploit code observed for this vulnerability.
9 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
A Django/GeoDjango vulnerability in GDAL raster handling that can allow attacker-controlled file write and outbound network requests via spatial lookups, potentially leading to remote code execution if a written file is later imported by the application.
A high-severity Django vulnerability in spatial lookups that allowed attacker-controlled str or dict values to reach GDALRaster, potentially causing server-side file writes and server-side request forgery, and in some cases enabling remote code execution depending on the raster driver.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.