PyTorch — CVE history (PyPI)
PyTorch
32 CVEs affect the PyTorch PyPI package (highest CVSS 9.8). Latest disclosed: 2026-03-22. Full CVE history sourced from NVD.
Summary
- Package
PyTorch(PyPI)- Total CVEs
32- Actively exploited (CISA KEV)
- 0
- Highest CVSS
9.8- Latest disclosed
- 2026-03-22
Recent CVEs (top 20)
| CVE | Severity | CVSS | KEV | Published | Summary |
|---|---|---|---|---|---|
CVE-2026-4538 | Medium | 5.3 | — | 2026-03-22 | A vulnerability was identified in PyTorch 2.10.0. |
CVE-2026-24747 | High | 8.8 | — | 2026-01-27 | PyTorch is a Python package that provides tensor computation. |
CVE-2025-63396 | Low | 3.3 | — | 2025-11-12 | An issue was discovered in PyTorch v2.5 and v2.7.1. |
CVE-2025-55560 | High | 7.5 | — | 2025-09-25 | An issue in pytorch v2.7.0 can lead to a Denial of Service (DoS) when a PyTorch model consists of torch.Tensor.to_sparse() and torch.Tensor.to_dense() and is compiled by Inductor. |
CVE-2025-55558 | High | 7.5 | — | 2025-09-25 | A buffer overflow occurs in pytorch v2.7.0 when a PyTorch model consists of torch.nn.Conv2d, torch.nn.functional.hardshrink, and torch.Tensor.view-torch.mv() and is compiled by Inductor, leading to a Denial of Service (DoS). |
CVE-2025-55557 | High | 7.5 | — | 2025-09-25 | A Name Error occurs in pytorch v2.7.0 when a PyTorch model consists of torch.cummin and is compiled by Inductor, leading to a Denial of Service (DoS). |
CVE-2025-55554 | Medium | 5.3 | — | 2025-09-25 | pytorch v2.8.0 was discovered to contain an integer overflow in the component torch.nan_to_num-.long(). |
CVE-2025-55553 | High | 7.5 | — | 2025-09-25 | A syntax error in the component proxy_tensor.py of pytorch v2.7.0 allows attackers to cause a Denial of Service (DoS). |
CVE-2025-55552 | High | 7.5 | — | 2025-09-25 | pytorch v2.8.0 was discovered to display unexpected behavior when the components torch.rot90 and torch.randn_like are used together. |
CVE-2025-55551 | High | 7.5 | — | 2025-09-25 | An issue in the component torch.linalg.lu of pytorch v2.8.0 allows attackers to cause a Denial of Service (DoS) when performing a slice operation. |
CVE-2025-46153 | Medium | 5.3 | — | 2025-09-25 | PyTorch before 3.7.0 has a bernoulli_p decompose function in decompositions.py even though it lacks full consistency with the eager CPU implementation, negatively affecting nn.Dropout1d, nn.Dropout2d, and nn.Dropout3d for fallback_random=T… |
CVE-2025-46152 | Medium | 5.3 | — | 2025-09-25 | In PyTorch before 2.7.0, bitwise_right_shift produces incorrect output for certain out-of-bounds values of the "other" argument. |
CVE-2025-46150 | Medium | 5.3 | — | 2025-09-25 | In PyTorch before 2.7.0, when torch.compile is used, FractionalMaxPool2d has inconsistent results. |
CVE-2025-46149 | Medium | 5.3 | — | 2025-09-25 | In PyTorch before 2.7.0, when inductor is used, nn.Fold has an assertion error. |
CVE-2025-46148 | Medium | 5.3 | — | 2025-09-25 | In PyTorch through 2.6.0, when eager is used, nn.PairwiseDistance(p=2) produces incorrect results. |
CVE-2025-4287 | Low | 3.3 | — | 2025-05-05 | A vulnerability was found in PyTorch 2.6.0+cu124. |
CVE-2025-32434 | Critical | 9.8 | — | 2025-04-18 | PyTorch is a Python package that provides tensor computation with strong GPU acceleration and deep neural networks built on a tape-based autograd system. |
CVE-2025-3730 | Low | 3.3 | — | 2025-04-16 | A vulnerability, which was classified as problematic, was found in PyTorch 2.6.0. |
CVE-2025-3136 | Low | 3.3 | — | 2025-04-03 | A vulnerability, which was classified as problematic, has been found in PyTorch 2.6.0. |
CVE-2025-3121 | Low | 3.3 | — | 2025-04-02 | A vulnerability classified as problematic has been found in PyTorch 2.6.0. |
All-time worst (top 10 by CVSS)
| CVE | Severity | CVSS | KEV | Published | Summary |
|---|---|---|---|---|---|
CVE-2025-32434 | Critical | 9.8 | — | 2025-04-18 | PyTorch is a Python package that provides tensor computation with strong GPU acceleration and deep neural networks built on a tape-based autograd system. |
CVE-2024-48063 | Critical | 9.8 | — | 2024-10-29 | In PyTorch <=2.4.1, the RemoteModule has Deserialization RCE. |
CVE-2022-45907 | Critical | 9.8 | — | 2022-11-26 | In PyTorch before trunk/89695, torch.jit.annotations.parse_type_line can cause arbitrary code execution because eval is used unsafely. |
CVE-2026-24747 | High | 8.8 | — | 2026-01-27 | PyTorch is a Python package that provides tensor computation. |
CVE-2024-31583 | High | 7.8 | — | 2024-04-17 | Pytorch before version v2.2.0 was discovered to contain a use-after-free vulnerability in torch/csrc/jit/mobile/interpreter.cpp. |
CVE-2025-55560 | High | 7.5 | — | 2025-09-25 | An issue in pytorch v2.7.0 can lead to a Denial of Service (DoS) when a PyTorch model consists of torch.Tensor.to_sparse() and torch.Tensor.to_dense() and is compiled by Inductor. |
CVE-2025-55558 | High | 7.5 | — | 2025-09-25 | A buffer overflow occurs in pytorch v2.7.0 when a PyTorch model consists of torch.nn.Conv2d, torch.nn.functional.hardshrink, and torch.Tensor.view-torch.mv() and is compiled by Inductor, leading to a Denial of Service (DoS). |
CVE-2025-55557 | High | 7.5 | — | 2025-09-25 | A Name Error occurs in pytorch v2.7.0 when a PyTorch model consists of torch.cummin and is compiled by Inductor, leading to a Denial of Service (DoS). |
CVE-2025-55553 | High | 7.5 | — | 2025-09-25 | A syntax error in the component proxy_tensor.py of pytorch v2.7.0 allows attackers to cause a Denial of Service (DoS). |
CVE-2025-55552 | High | 7.5 | — | 2025-09-25 | pytorch v2.8.0 was discovered to display unexpected behavior when the components torch.rot90 and torch.randn_like are used together. |