OpenSSL CVE-2026-121481 flaw rated 9.8 but Low risk

A flaw in OpenSSL has been given the highest possible severity score of 9.8 out of 10, yet the OpenSSL project itself rates the issue Low. The disagreement, which played out in public on 20 November 2026, is a reminder that a severity score alone does not tell administrators how urgently to act.
The flaw, tracked as CVE-2026-121481, is an integer overflow in EVP_EncodeBlock(), the function OpenSSL uses to base64-encode a buffer in a single call. It was published on 18 November 2026, and the CVE record was updated on 20 November.
A 9.8 that the vendor disputes
The 9.8 score comes from CISA's vulnerability enrichment programme, which attached a CVSS 3.1 vector of AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H to the flaw. The US National Vulnerability Database still lists the record as awaiting analysis, and carries only that secondary score.
OpenSSL, which is the numbering authority for its own CVEs, gives no CVSS score for the flaw and rates it Low under its security policy. Its impact summary states plainly that no path by which an attacker can trigger the flaw has been identified. On 20 November, an OpenSSL committer asked CISA to rescore the flaw, arguing that the length involved comes from the calling application, not from an attacker.
How the overflow works
EVP_EncodeBlock() takes a buffer and a length supplied by the caller, and writes the base64 encoding into an output buffer the caller allocates. The caller is expected to reserve 4 times the length plus two, divided by three, plus one byte. When the input length is close to INT_MAX, the largest value a signed 32-bit integer can hold, the arithmetic that computes the encoded size overflows, and the function writes more data than the documented output size.
The result is a heap buffer overflow in the calling process, which could crash it or corrupt its memory. Reaching the flaw requires an application to pass an input length near INT_MAX to the function, which means it must already hold a buffer of about two gigabytes. OpenSSL's own TLS, QUIC and CMS code never encodes buffers of that size, and its command-line tools cannot practically be made to do so. An application that accepts an attacker-controlled length and forwards it to EVP_EncodeBlock() without its own bound check would be exposed, but no such application has been named.
Who is affected
Only two OpenSSL release lines are affected: 4.0.0 up to but not including 4.0.4, and 4.1.0 up to but not including 4.1.1. In shipped software that means 4.0.0 through 4.0.3, plus 4.1.0. OpenSSL 3.x is outside the affected set. The weakness is classified as CWE-190, integer overflow or wraparound, with the resulting heap overflow classified as CWE-122.
Ubuntu has marked the issue as medium priority, and notes that its OpenSSL package still needs evaluation. The flaw is not listed in CISA's Known Exploited Vulnerabilities catalog, and no in-the-wild exploitation, public proof of concept or attribution has been reported.
When the fix arrives
There is no fixed release to download yet. The fix exists as a public pull request, number 33652, and the releases that will carry it, 4.0.4 and 4.1.1, are scheduled for 1 December 2026, alongside fixes from the project's 10 and 17 November advisories.
That leaves administrators in an unusual position: the correct advice is to upgrade to 4.0.4 or 4.1.1, but neither version exists yet. Until then, the practical step is to check whether any application in use passes very large, attacker-influenced lengths to EVP_EncodeBlock(), and to apply the December releases promptly when they arrive. For most users, whose software never encodes buffers anywhere near two gigabytes, the flaw is a low-priority item despite its headline score.