NumPy vulnerabilities: known CVEs & security history
NumPy · Library / Python · 8 tracked CVEs · 0 actively exploited · updated June 2026 · what is a CVE? →
This is the full list of known vulnerabilities (CVEs) across all NumPy release lines — 8 in total. A CVE here doesn't mean your version is affected — check NumPy's current status and the safe version to run.
Known NumPy CVEs
Actively-exploited and most-severe first. Open any CVE for full details.
| CVE | Severity | CVSS | EPSS | Year |
|---|---|---|---|---|
| CVE-2019-6446 | critical | 9.8 | 17% | 2019 |
| CVE-2017-12852 | high | 7.5 | 3% | 2017 |
| CVE-2021-41496 | medium | 5.5 | 0% | 2021 |
| CVE-2014-1859 | medium | 5.5 | 0% | 2018 |
| CVE-2014-1858 | medium | 5.5 | 0% | 2018 |
| CVE-2021-41495 | medium | 5.3 | 1% | 2021 |
| CVE-2021-34141 | medium | 5.3 | 2% | 2021 |
| CVE-2021-33430 | medium | 5.3 | 1% | 2021 |
Is my NumPy version affected?
The list above spans every release. To know whether your version is affected — and the minimum safe version to upgrade to — check it directly.
Check your NumPy version → · Monitor NumPy for new CVEs →
NumPy vulnerabilities — frequently asked
How many known vulnerabilities does NumPy have?
IsItPatched tracks 8 CVEs for NumPy. 1 is critical-severity and 1 high-severity. These span every release line — what matters is whether the version you run is affected.
Does NumPy have any actively-exploited vulnerabilities?
None of NumPy's tracked CVEs are currently in CISA's KEV catalog — but new ones can be added at any time, so keep your version current.
What is the most severe NumPy vulnerability?
Among tracked issues, CVE-2019-6446 (CRITICAL, CVSS 9.8) ranks highest — a Insecure deserialization weakness.
Is NumPy safe to use?
It depends on the version. The latest supported NumPy release (2.4.6) clears the known issues; older versions may still be affected. Check the exact version you run for a verdict.
CVE data aggregated from NVD, CISA KEV and EPSS (FIRST.org). Related: NumPy security status · NumPy end-of-life · actively-exploited CVEs. Always verify against NumPy's advisories — see our disclaimer.