{"description":"Trending threats, MITRE ATT\u0026CK coverage, and detection metadata. Fed continuously.","feed_url":"https://feed.craftedsignal.io/products/torchvision-through-0.28.0/feed.json","home_page_url":"https://feed.craftedsignal.io/","items":[{"_cs_actors":[],"_cs_cpes":[],"_cs_cves":[{"cvss":7.1,"id":"CVE-2026-65918"}],"_cs_exploited":false,"_cs_has_poc":false,"_cs_poc_references":[],"_cs_products":["torchvision (through 0.28.0)"],"_cs_severities":["high"],"_cs_tags":["denial-of-service","information-disclosure","library-vulnerability"],"_cs_type":"advisory","_cs_vendors":["PyTorch"],"content_html":"\u003cp\u003eA critical out-of-bounds heap read vulnerability, identified as CVE-2026-65918, affects PyTorch's \u003ccode\u003etorchvision\u003c/code\u003e library up to and including version 0.28.0. This flaw resides within the GIF decoder's \u003ccode\u003eread_from_tensor\u003c/code\u003e callback, which inadvertently passes an unchecked (unclamped) length value to the \u003ccode\u003ememcpy\u003c/code\u003e function. This oversight enables attackers to craft and supply malicious or truncated GIF files. When a vulnerable application processes these files, the uncontrolled memory access can lead to a denial of service (DoS) through a segmentation fault, causing the application to crash. Furthermore, the vulnerability could potentially allow for the disclosure of sensitive data residing in adjacent heap memory regions. The issue was addressed in commit \u003ccode\u003e4e05dc2\u003c/code\u003e. This vulnerability is particularly concerning for applications that process untrusted image inputs, as successful exploitation could lead to application instability or sensitive information exposure.\u003c/p\u003e\n\u003ch2 id=\"attack-chain\"\u003eAttack Chain\u003c/h2\u003e\n\u003col\u003e\n\u003cli\u003e\u003cstrong\u003eAttacker crafts malicious GIF\u003c/strong\u003e: An attacker develops a specially crafted or truncated GIF file designed to exploit the out-of-bounds read vulnerability in the \u003ccode\u003ePyTorch torchvision\u003c/code\u003e library.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eDelivery of malicious GIF\u003c/strong\u003e: The attacker delivers the malicious GIF to a target system. This could occur via various vectors, such as an email attachment, a malicious website download, or embedded within a web application where \u003ccode\u003etorchvision\u003c/code\u003e processes user-supplied images.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eProcessing by vulnerable application\u003c/strong\u003e: A software application utilizing the vulnerable \u003ccode\u003ePyTorch torchvision\u003c/code\u003e library (version 0.28.0 or earlier) attempts to process the malicious GIF file for tasks like display, analysis, or transformation.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003e\u003ccode\u003eread_from_tensor\u003c/code\u003e callback execution\u003c/strong\u003e: During the GIF decoding process, the \u003ccode\u003eread_from_tensor\u003c/code\u003e callback function within \u003ccode\u003etorchvision\u003c/code\u003e is invoked to handle the image data from the malicious file.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eOut-of-bounds read via \u003ccode\u003ememcpy\u003c/code\u003e\u003c/strong\u003e: Within the \u003ccode\u003eread_from_tensor\u003c/code\u003e callback, an unclamped (unchecked) length value is passed to the \u003ccode\u003ememcpy\u003c/code\u003e function, triggering an attempt to read data beyond the boundaries of the allocated heap buffer.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eDenial of Service (DoS)\u003c/strong\u003e: The out-of-bounds read operation results in a segmentation fault, causing the application to crash and rendering it unavailable, leading to a denial of service.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eInformation Disclosure\u003c/strong\u003e: In some scenarios, instead of a crash, the out-of-bounds read may disclose sensitive data from adjacent heap memory regions. This leaked information could then be potentially exfiltrated or leveraged by the attacker for further exploitation.\u003c/li\u003e\n\u003c/ol\u003e\n\u003ch2 id=\"impact\"\u003eImpact\u003c/h2\u003e\n\u003cp\u003eSuccessful exploitation of CVE-2026-65918 can lead to a high impact on the availability and confidentiality of systems using vulnerable versions of \u003ccode\u003ePyTorch torchvision\u003c/code\u003e. The primary observed damage includes denial of service, characterized by application crashes due to segmentation faults, disrupting normal operations. Depending on the memory layout and the nature of the data, exploitation could also lead to the disclosure of adjacent heap memory contents, potentially exposing sensitive information processed by the application. While the NVD does not specify victim counts or targeted sectors, any application that processes untrusted GIF files using \u003ccode\u003ePyTorch torchvision\u003c/code\u003e through version 0.28.0 is at risk.\u003c/p\u003e\n\u003ch2 id=\"recommendation\"\u003eRecommendation\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003ePatch CVE-2026-65918 immediately\u003c/strong\u003e: Update \u003ccode\u003ePyTorch torchvision\u003c/code\u003e to a version beyond 0.28.0 that includes commit \u003ccode\u003e4e05dc22f5f050a9528cc0ea09ceca6cdaf8f4ed\u003c/code\u003e to remediate the vulnerability.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eMonitor for application crashes\u003c/strong\u003e: Implement monitoring for unexpected process terminations or segmentation faults in applications utilizing \u003ccode\u003ePyTorch torchvision\u003c/code\u003e as a potential indicator of attempted exploitation, even though this is not specific to CVE-2026-65918.\u003c/li\u003e\n\u003c/ul\u003e\n","date_modified":"2026-07-23T18:24:31Z","date_published":"2026-07-23T18:24:31Z","id":"https://feed.craftedsignal.io/briefs/2026-07-pytorch-torchvision-oob-read/","summary":"An out-of-bounds heap read vulnerability (CVE-2026-65918) in PyTorch torchvision through version 0.28.0 allows attackers to supply malicious GIF files, leading to denial of service via segmentation fault or disclosure of adjacent heap memory contents.","title":"CVE-2026-65918 - PyTorch Torchvision Out-of-Bounds Heap Read Vulnerability","url":"https://feed.craftedsignal.io/briefs/2026-07-pytorch-torchvision-oob-read/"}],"language":"en","title":"CraftedSignal Threat Feed - Torchvision (Through 0.28.0)","version":"https://jsonfeed.org/version/1.1"}