CVE-2026-15976

CVE-2026-15976 is a critical-severity vulnerability in Lmsys Sglang with a CVSS 3.x base score of 9.8. It is not currently listed as actively exploited by CISA, and its EPSS exploit-prediction score is low. The underlying weakness is classified as CWE-502.

Key facts

Description

SGLang contains a RCE vulnerability when attempting to load model weights from a HuggingFace repository, specifically within the /update_weights_from_disk, where torch.load(..., weights_only=False) fallback enables pickle deserialization of .bin files.

Frequently asked questions

What is CVE-2026-15976?
SGLang contains a RCE vulnerability when attempting to load model weights from a HuggingFace repository, specifically within the /update_weights_from_disk, where torch.load(..., weights_only=False) fallback enables pickle deserialization of .bin files.
How severe is CVE-2026-15976?
CVE-2026-15976 has a CVSS 3.x base score of 9.8, rated critical severity. It is exploitable over network with low attack complexity, requires no privileges and no user interaction. Impact on confidentiality is high, integrity high, and availability high.
Is CVE-2026-15976 being actively exploited?
It is not currently listed in CISA's KEV catalog. Its EPSS exploit-prediction score is 0% (26th percentile), an estimate of the probability of exploitation in the next 30 days.
What products are affected by CVE-2026-15976?
CVE-2026-15976 affects Lmsys Sglang. See the affected-products list for the exact vulnerable versions.
How do I fix CVE-2026-15976?
Review the linked vendor and NVD advisories for patched versions and mitigations, then upgrade or apply the recommended workaround. Given its critical severity, prioritise patching exposed systems.
When was CVE-2026-15976 published?
CVE-2026-15976 was published on 2026-07-30 and last updated on 2026-08-04.

References

Affected products (1)

More vulnerabilities in Lmsys Sglang

All CVEs affecting Lmsys Sglang →

Other CWE-502 (Deserialization of Untrusted Data) vulnerabilities

Browse all CWE-502 (Deserialization of Untrusted Data) vulnerabilities →