CVE-2026-94418 is a certificate signature bypass in low-memory builds of the wolfSSL embedded TLS library: with WOLFSSL_SMALL_CERT_VERIFY, a parse error hid a failed signature check, so an application whose verify callback overrides date errors could accept a self-made certificate. NVD scores it 7.5, wolfSSL rates it Low, and wolfSSL 5.9.4 fixes it.
Vulnerability summary
CVSS base score
7.5
High severity
EPSS score (FIRST)
0.1%
Probability of exploitation in the wild in the next 30 days
CISA KEV catalog
No
Not listed
Published
Vulnerability ID: CVE-2026-94418 · Source: NVD
Affected components
| Component | CPE | Version range |
|---|---|---|
| wolfSSL (WOLFSSL_SMALL_CERT_VERIFY peer certificate checks) | cpe:2.3:a:wolfssl:wolfssl:*:*:*:*:*:*:*:* | NVD: 3.15.5 through 5.9.2; only WOLFSSL_SMALL_CERT_VERIFY builds (--enable-lowresource, --enable-leantls, --enable-tinytls13=cert or mutualauth) with a date-override verify callback; fixed in 5.9.4 (PR 11500) |
What does this mean for affected devices?
CVE-2026-94418 is an improper verification of a cryptographic signature (CWE-347, per wolfSSL as CNA). The 5.9.4 section of the wolfSSL ChangeLog explains that under WOLFSSL_SMALL_CERT_VERIFY, ProcessPeerCertParse() runs the signature check separately "to keep peak memory down" but merged its result back "only when the parse returned 0, so any parse error hid it". The attacker "needs no key material from the real PKI and no CA compromise": a self-made certificate with the expected subject, the trusted CA's subject as issuer, arbitrary signature bytes, a validity window in the past and the attacker's own key pair is enough. Releases 3.15.5 through 5.9.2 are affected; the issue was found via the Anthropic OSS program and fixed in pull request 11500.
Two conditions must both hold. First, the build defines WOLFSSL_SMALL_CERT_VERIFY, which wolfSSL says is off by default and reached through --enable-lowresource, --enable-leantls, --enable-tinytls13=cert and --enable-tinytls13=mutualauth, or through WC_CFG_SMALL_CERT_VERIFY in examples/configs/user_settings_embedded.h, which ships as 0; --enable-all and --enable-distro never set it. In wolfSSL's configure.ac the low-resource option, described as "low resource options to reduce flash and memory use", adds -DWOLFSSL_SMALL_CERT_VERIFY. Second, the application installs a verify callback that returns 1 for ASN_BEFORE_DATE_E or ASN_AFTER_DATE_E; wolfSSL ships that shape as myVerify() in wolfssl/test.h. Both point to constrained devices: the build options exist to save flash and RAM, and overriding date errors is a pattern for a device that cannot trust its own clock.
NVD gives it a Primary CVSS v3.1 score of 7.5 with vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N and covers 3.15.5 through 5.9.2; wolfSSL's own CVSS v4.0 score is 2.3 and the wolfSSL vulnerabilities page rates it Low (NVD). TLS 1.2, TLS 1.3 and DTLS are affected in both directions, and where the forged certificate is a chain certificate it is cached in the WOLFSSL_CTX certificate manager, so an exposed device must restart the context or the process rather than just reconnect.
The Debian security tracker marks the wolfssl package "unimportant" and bookworm as ignored because "WOLFSSL_SMALL_CERT_VERIFY not built", which shows the split: general-purpose distribution packages sit outside the condition, while custom low-memory firmware builds can sit inside it.
CVE-2026-94418 is not in the CISA KEV catalog as of 2026-11-16 (catalog version 2026.11.13), and FIRST EPSS gives it 0.0011, the 1.2nd percentile, in the 2026-11-15 dataset (CISA KEV, FIRST EPSS).
Search the firmware build for WOLFSSL_SMALL_CERT_VERIFY, the low-resource configure options and any verify callback that accepts date errors; if all are present, update to 5.9.4. Our Mbed TLS and wolfSSL firmware guide covers reading user_settings.h and configure lines. Other wolfSSL certificate flaws are CVE-2026-93302 and CVE-2026-5194.
Evidence
- Under WOLFSSL_SMALL_CERT_VERIFY a parse error hid the signature result; self-made certificate suffices; build routes --enable-lowresource, --enable-leantls, --enable-tinytls13=cert/mutualauth, user_settings_embedded.h WC_CFG_SMALL_CERT_VERIFY (ships 0); not set by --enable-all/--enable-distro; requires a date-override verify callback (myVerify in wolfssl/test.h); releases 3.15.5 through 5.9.2; Anthropic OSS program; fixed in PR 11500, release 5.9.4. — wolfSSL ChangeLog (5.9.4)
- Low rating, fixed in 5.9.4. — wolfSSL security vulnerabilities page
- NVD Primary CVSS v3.1 7.5 (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N); CNA CVSS v4.0 2.3; CPE range 3.15.5 through 5.9.2. — NVD
- --enable-lowresource ("low resource options to reduce flash and memory use") adds -DWOLFSSL_SMALL_CERT_VERIFY. — wolfSSL configure.ac (v5.9.2-stable)
- PR 11500 fixes the WOLFSSL_SMALL_CERT_VERIFY signature merge. — wolfSSL pull request 11500
- Debian: unimportant; bookworm ignored, WOLFSSL_SMALL_CERT_VERIFY not built. — Debian security tracker
- Not in the CISA KEV catalog as of 2026-11-16 (catalog version 2026.11.13). — CISA KEV
- EPSS 0.00107 (1.2nd percentile), dataset 2026-11-15, read 2026-11-16. — FIRST EPSS

Related advisories
Sources
Frequently asked questions
Which wolfSSL builds are affected by CVE-2026-94418?
Builds that define WOLFSSL_SMALL_CERT_VERIFY, reached through --enable-lowresource, --enable-leantls or --enable-tinytls13=cert or mutualauth, and whose application verify callback returns 1 for date errors. The macro is off by default and --enable-all and --enable-distro do not set it (per the wolfSSL ChangeLog).
What does an attacker need to exploit CVE-2026-94418?
No CA compromise or real key material: a self-made certificate with the expected subject, the trusted CA's subject as issuer, arbitrary signature bytes, a past validity window and the attacker's own key pair (per the wolfSSL ChangeLog).
Are Linux distribution packages affected by CVE-2026-94418?
Debian marks it unimportant and ignores it for bookworm because WOLFSSL_SMALL_CERT_VERIFY is not built there; the exposure is in custom low-memory builds (per the Debian security tracker).
Is CVE-2026-94418 exploited in the wild?
It is not in the CISA KEV catalog as of 2026-11-16 (catalog version 2026.11.13), and FIRST EPSS gives it 0.0011, the 1.2nd percentile, in the 2026-11-15 dataset (per CISA KEV and FIRST EPSS).
Questions about this page?