CVE-2026-65918 PyTorch torchvision GIF Decoder Out-of-bounds Heap Read
Attackers can cause denial of service or disclose heap memory by supplying malicious GIF files to the PyTorch torchvision library.
- Exploited in the wild
- Not known
- Login needed
- Not needed
- Attackable
- Over the internet
- Victim action needed
- A user has to click or open something
- Fix
- None listed
What is CVE-2026-65918?
CVE-2026-65918 is a high-severity vulnerability in pytorch vision, published on 23 Jul 2026. An attacker can exploit it over the internet without logging in, and only if a user opens a file or follows a link.
Official description
PyTorch torchvision through 0.28.0, fixed in commit 4e05dc2, contains an out-of-bounds heap read vulnerability in the GIF decoder's read_from_tensor callback that passes unclamped length to memcpy. Attackers can supply malicious or truncated GIF files to cause denial of service via segmentation fault or disclose adjacent heap memory contents.
Weakness: CWE-125
Which versions are affected by CVE-2026-65918?
It affects 1 product.
Products include Linuxfoundation Torchvision.
How do I fix CVE-2026-65918?
No fix has been published for CVE-2026-65918 yet. Limit exposure of the affected software until one is available.
Is CVE-2026-65918 being exploited?
It is not confirmed as exploited in the wild, but public exploit code for CVE-2026-65918 exists.
Sources
Also known as DEBIAN-CVE-2026-65918, GHSA-xhgh-4jhq-chj3, UBUNTU-CVE-2026-65918.