{"action":"Upgrade torch to 2.7.1 or later.","affected":[{"fixed":"2.7.1","introduced":"0"}],"component":"torch","cve_ids":["CVE-2025-55558"],"ecosystem":"PyPI","epss":0.0045,"exploited":false,"fixed_versions":["2.7.1"],"id":"tcs_01M5KGP6VEH1ASWDQK7SHPB2SY","modified_at":"2026-05-20T09:19:20.054109Z","published_at":"2025-09-25T16:15:34.960Z","references":["https://osv.dev/vulnerability/PYSEC-2025-208","https://gist.github.com/shaoyuyoung/0e7d2a586297ae9c8ed14d8706749efc","https://github.com/pytorch/pytorch/issues/151523","https://github.com/pytorch/pytorch/pull/151887"],"severity":{"label":"high","score":7.5},"sources":[{"source":"osv","source_id":"PYSEC-2025-208","url":"https://osv.dev/vulnerability/PYSEC-2025-208"}],"summary":"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"}