CVE-2023-2283 is an authentication bypass in libssh 0.9 and 0.10 with the OpenSSL backend: when a memory allocation fails during signature checking, pki_verify_data_signature returns success, so an invalid signature can pass. NVD scores it 6.5; 0.9.7 and 0.10.5 fix it, and OpenWrt 23.05 still packages 0.10.4 with OpenSSL.
Vulnerability summary
CVSS base score
6.5
Medium severity
EPSS score (FIRST)
1.1%
Probability of exploitation in the wild in the next 30 days
CISA KEV catalog
No
Not listed
Published
Vulnerability ID: CVE-2023-2283 · Source: NVD
Affected components
| Component | CPE | Version range |
|---|---|---|
| libssh (OpenSSL backend) | cpe:2.3:a:libssh:libssh:*:*:*:*:*:*:*:* | 0.9.0 to 0.9.6 and 0.10.0 to 0.10.4 per libssh (NVD: 0.9.1 to 0.9.6, 0.10.0 to 0.10.4); fixed in 0.9.7 and 0.10.5 |
| OpenWrt packages feed, openwrt-23.05 branch: libssh | cpe:/a:libssh:libssh | 0.10.4 built with libopenssl, WITH_SERVER and WITH_SFTP on, no patches (package Makefile read 2026-11-16) |
What does this mean for affected devices?
CVE-2023-2283 is an improper authentication flaw (CWE-287) in libssh's pki_verify_data_signature(). The libssh advisory explains that the return value is overwritten by an earlier call and "any error between them calls goto error returning SSH_OK", which can happen when there is not enough memory or memory usage is limited. Only the OpenSSL cryptographic backend is affected, and libssh 0.9.7 and 0.10.5 fix it; the advisory's workaround is to build with the libgcrypt or mbedTLS backend.
NVD scores it 6.5 (medium) with vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N and lists libssh 0.9.1 through 0.9.6 and 0.10.0 through 0.10.4 (NVD), while the libssh advisory itself names 0.9.0 to 0.9.6 and 0.10.0 to 0.10.4 and scores it 4.2 under CVSS 3.0. The trigger, a failed allocation, is a more realistic condition on a small device with tight memory limits than on a server.
The firmware angle is the build-system package. On 2026-11-16 the OpenWrt packages Makefile for the openwrt-23.05 branch builds libssh 0.10.4, links it against libopenssl, turns the server and SFTP code on (WITH_SERVER and WITH_SFTP) and carries no patches, so an image built from that branch with libssh has the affected version and backend. The master branch switched libssh to mbedTLS on 2023-08-28 and has since moved to the 0.12 series. Our libssh and libssh2 vulnerabilities in embedded firmware guide shows how to find the libssh version, backend and server role in an image.
The CVE is not in the CISA KEV catalog as of 2026-11-16 (catalog version 2026.11.13), and FIRST EPSS gives it 0.011, the 65th percentile, in the 2026-11-15 dataset (CISA KEV, FIRST EPSS). The Debian tracker shows bullseye fixed in 0.9.7-0+deb11u1 (DSA-5409-1) and buster not affected because the vulnerable code arrived later.
Device makers should move to libssh 0.10.5 or later, or rebuild with the mbedTLS or libgcrypt backend, and treat this as a server-side issue only where the product accepts SSH logins through libssh. Our authentication bypass fix guide covers testing that failure paths in authentication code deny access.
Evidence
- pki_verify_data_signature can accept invalid signatures when memory allocation fails; OpenSSL backend only; fixed in 0.9.7 and 0.10.5; workaround libgcrypt or mbedTLS backend. — libssh security advisory
- NVD CVSS v3.1 6.5; CWE-287; affected 0.9.1 to 0.9.6 and 0.10.0 to 0.10.4. — NVD
- openwrt-23.05 libssh package: PKG_VERSION 0.10.4, DEPENDS libopenssl, WITH_SERVER and WITH_SFTP on, Makefile only. — OpenWrt packages (openwrt-23.05)
- OpenWrt master switched libssh to MbedTLS on 2023-08-28. — OpenWrt packages commit history
- Not in the CISA KEV catalog as of 2026-11-16 (catalog version 2026.11.13). — CISA KEV
- EPSS 0.01061 (65th percentile), dataset 2026-11-15, read 2026-11-16. — FIRST EPSS

Related advisories
Sources
Frequently asked questions
Is every libssh build affected by CVE-2023-2283?
No. The libssh advisory states only the OpenSSL cryptographic backend is affected, in 0.9.0 to 0.9.6 and 0.10.0 to 0.10.4; builds using libgcrypt or mbedTLS are outside it (per the libssh advisory).
Why does CVE-2023-2283 matter more on embedded devices?
The bypass needs a memory allocation to fail during signature checking, which the libssh advisory says may happen when there is not enough memory or memory usage is limited, a common state on small devices (per the libssh advisory).
Does OpenWrt ship an affected libssh?
On 2026-11-16 the openwrt-23.05 packages branch builds libssh 0.10.4 against OpenSSL with server and SFTP support and no patches; newer branches use mbedTLS and the 0.12 series (per the OpenWrt packages repository).
Is CVE-2023-2283 in the CISA KEV catalog?
No, not as of 2026-11-16 (catalog version 2026.11.13). FIRST EPSS gives it 0.011, the 65th percentile, in the 2026-11-15 dataset (per CISA KEV and FIRST EPSS).
Questions about this page?