PostgreSQL type confusion ALTER SET DATA TYPE Flaw CVE-2026-6460: Update Now
Published November 12, 2026 · Last updated November 12, 2026
The short version: CVE-2026-6460 is a high-severity security flaw in PostgreSQL type confusion ALTER SET DATA TYPE. The National Vulnerability Database (NVD) published it on November 12, 2026, and the score it lists is 8.8 out of 10. When FraudMeNot checked on November 12, 2026, CISA had not added CVE-2026-6460 to its Known Exploited Vulnerabilities catalog, so there was no public U.S. government confirmation of attacks yet. A fix is available, which is why this page focuses on who should update and how.
CVE-2026-6460 is a high-severity type-confusion flaw in the PostgreSQL core server. The PostgreSQL Global Development Group (PGDG) says an object creator can execute arbitrary code as the operating-system user that runs the database, primarily through ALTER ... SET DATA TYPE. Other data-definition routes exist, including ALTER TABLE ... ADD COLUMN; ROLLBACK; ALTER TABLE ... ADD COLUMN. The project’s own description is that the server “had inadequate tracking of typed values, and values were not self-identifying,” so stale values could appear in data-dictionary entries, PL/pgSQL variables, and many other parts of the system.
PGDG’s November 12, 2026 announcement lists this issue first among the security fixes and scores it CVSS v3.1 8.8. The dedicated advisory page labels the same figure “CVSS 3.0 Overall Score 8.8” with vector AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H (network, low complexity, low privileges, no user interaction, high impact on confidentiality, integrity, and availability). NVD records that vector as CVSS 3.1 8.8 HIGH from the PostgreSQL CNA, type Secondary; there is no NVD Primary and no CVSS 4.0. The selection rule prefers V31, then V40, then V30, so the selected score is CVSS 3.1 8.8. That meets the 7.0 floor. The vendor headline on the announcement is also CVSS v3.1 8.8.
The November 12 update
The same day, PGDG released PostgreSQL 18.7, 17.12, 16.16, 15.20, and 14.25, plus the second release candidate of PostgreSQL 19. The announcement is explicit: this update “fixes 52 security vulnerabilities and over 140 bugs reported over the last several months,” and “there are several issues that may require you to take extra steps after updating.”
Of those 52 advisories, 35 are scored at or above 7.0 on the announcement’s CVSS v3.1 figures, and a large share are 8.8—the same as the lead issue—covering use-after-free and type-confusion bugs that can run code as the database’s operating-system user. This page treats CVE-2026-6460 as the lead because PGDG lists it first and it is jointly the highest score. Installing the named minor releases addresses the whole set. The remaining high-severity items are summarized below, not drafted as separate pages.
The 18.7 release notes, dated November 12, say a dump and restore is not required for sites already on 18.X. They also warn that several bugs can leave indexes corrupt, that existing incremental backup files should not be trusted, and that the defense work for CVE-2026-6460 is a “voluminous set of patches.” The notes describe versioning logic so the server can detect composite values that no longer match the tuple descriptor they are interpreted with. For full protection, C-language extensions may need to be rebuilt, and some extensions need source changes. A new composite_type_mismatch server parameter can disable those checks; the notes say doing so “carries security risks.” The update also adds locking to block cases such as ALTER TYPE on a composite type used in a CREATE TABLE that has not committed yet, and it tightens when ALTER COLUMN TYPE is allowed if dependencies exist—for example, it is no longer allowed when a function’s default expression or a new-style SQL function body depends on the type.
Who should act, and what to install
Database administrators and operators of supported PostgreSQL deployments should install the fixed minor release for their major line: 18.7, 17.12, 16.16, 15.20, or 14.25. The announcement’s Updating section says all update releases are cumulative and, as with other minor releases, a dump and reload or pg_upgrade is not required: “you may simply stop PostgreSQL and update its binaries.” Downloads and the full release notes are on postgresql.org.
This is the final release of PostgreSQL 14. The project now treats 14 as end-of-life and says it will no longer receive security and bug fixes. Sites still running 14 in production are told to plan a move to a newer, supported version, in addition to installing 14.25 so they are current for this last cycle.
Post-update steps the project asks you to take
These steps come from the official Updating section and the 18.7 migration notes. They are not optional color; PGDG calls them out because several of the non-security bug fixes can leave existing structures wrong.
- If you suspect a GIN index was hit by one of the corruption issues fixed in this release, reindex it (
REINDEX INDEX). Two GIN problems are described in the notes: a vacuum that needs multiple index scans could miss removing some entries, and a posting-tree page deletion after an incomplete split could disconnect sibling pages so later searches miss rows. - If you use
pg_trgm, reindex any GiST indexes that use thegist_trgm_opsoperator class. Combination of index entries could lose a format flag; the fix does not repair existing indexes. - If you see “missing chunk number” errors when reading a table, reindex the table (which also reindexes its TOAST table). The notes describe a case where, after
REINDEX CONCURRENTLYrebuilt an index on a TOAST table, later updates could go wrong. - If you use incremental backups, take a new full backup after updating and use it as the basis for later incrementals. The project says existing incremental files should not be trusted.
- If you use partitioned tables, a query published in the announcement finds
NOT NULL NO INHERITconstraints on partitioned tables. Unwanted constraints can be dropped; constraints you want to keep should be repaired withALTER TABLE ... ALTER CONSTRAINT ... INHERIT, which fails if any partition already contains NULL values in that column. - If you have B-tree indexes on
bpcharcolumns declared asbpcharwithout a length, reindex them. Columns declared aschar(n)are not affected.
Users who skipped earlier minor releases may have additional post-update steps from those earlier notes.
Other high-severity issues closed in the same builds
The same binaries also close a long list of CVSS v3.1 8.8 (and a few 8.1 / 8.0 / 7.2) items. Official one-line descriptions include: use-after-free in parallel RECORD type identifiers (CVE-2026-116304); type confusion when a composite type changes during field assignment (CVE-2026-116306); several plperl and pltcl use-after-free and type-confusion bugs (CVE-2026-116307, 116309, 116319, 116320, 116323, 116326); heap buffer overflows in tsvector_delete() and ts_rank_cd() (CVE-2026-116315, 116317); pg_restore executing code because non-plain archives were not validated (CVE-2026-116314); SQL injection through a pg_dump not-null constraint name (CVE-2026-116318) and through reindexdb index-name quoting (CVE-2026-116325); pg_dump tar-format restore.sql failing to block psql command injection (CVE-2026-116330); security-restricted operations creating types in pg_temp that later queries in the session find (CVE-2026-116316); logical replication being able to write to subscriber system catalogs (CVE-2026-120145); dblink and subscriptions appending to arbitrary files (CVE-2026-14661, 8.1) and a dblink disconnect use-after-free (CVE-2026-120147); CREATE EXTENSION / ALTER EXTENSION locating the wrong object after a schema rename, which lets a role with database-level CREATE privilege run arbitrary SQL as the bootstrap superuser (CVE-2026-6467, vendor CVSS v3.1 7.2); and vacuumdb --missing-stats-only letting a database owner execute code as the vacuumdb user (CVE-2026-116313, 7.2).
Lower-scored items in the same announcement—row-security and view-visibility mistakes, a pre-authentication cancel request that can leak a spinlock, memory disclosure, and similar—are also closed by the same upgrades. They are listed on the November 12 news page and are not expanded here.
Score, window, and sources
Selected score for the lead: CVSS 3.1 8.8 HIGH, PostgreSQL CNA, NVD Secondary; vendor announcement headline CVSS v3.1 8.8. First official announcement: PGDG news post dated 2026-11-12, with “Fix Published 2026-11-12” on the CVE-2026-6460 advisory. The 72-hour window ends November 15, 2026. This slot is inside that window.
Sources: PostgreSQL news “PostgreSQL 18.7, 17.12, 16.16, 15.20, 14.25, and 19 RC 2 Released!” posted 2026-11-12; PostgreSQL advisory page for CVE-2026-6460; PostgreSQL 18.7 release notes (release date 2026-11-12); NVD record for CVE-2026-6460.
What the official record says about CVE-2026-6460
The record was filed by f86ef6dc-4d3a-42ad-8f28-e6d5547a5007. In its own words: "Type confusion in PostgreSQL allows an object creator to execute arbitrary code as the operating system user running the database, via ALTER ... SET DATA TYPE primarily. Other DDL attack routes exist, such as "ALTER TABLE ... ADD COLUMN; ROLLBACK; ALTER TABLE ... ADD COLUMN". The system had inadequate tracking of typed values, and values were not self-identifying. Stale values could appear in data dictionary entries, PL/pgSQL variables, and many other parts of the system. PostgreSQL Versions before PostgreSQL 18.7, 17.12, 16.16, 15.20, and 14.25 are affected."
Why type confusion ALTER SET DATA TYPE admins should care
Severity scores use the Common Vulnerability Scoring System (CVSS). f86ef6dc-4d3a-42ad-8f28-e6d5547a5007 rated CVE-2026-6460 8.8 (High) under CVSS 3.1. Reading that rating in everyday terms: it can be reached over a network, including the internet; an attacker needs a low-privilege account; no one has to click or open anything. If it is abused, the rated damage to keeping data private is high, to keeping data unaltered is high, and to keeping type confusion ALTER SET DATA TYPE running is high. Scoring string: CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H.
Affected type confusion ALTER SET DATA TYPE releases
According to the record, these releases are vulnerable:
- 18 up to, but not including, 18.7
- 17 up to, but not including, 17.12
- 16 up to, but not including, 16.16
- 15 up to, but not including, 15.20
- versions before 14.25
The type of bug behind CVE-2026-6460
NVD classifies it using MITRE's Common Weakness Enumeration:
- CWE-843, Access of Resource Using Incompatible Type. MITRE describes this class of bug as follows: "The product allocates or initializes a resource such as a pointer, object, or variable using one type, but it later accesses that resource using a type that is incompatible with the original type. (Roles: Common-Name, Common-Name)"
Vendor remediation for CVE-2026-6460
The vendor advisory remediation for operators of type confusion ALTER SET DATA TYPE, quoted for this page, is: "Upgrade PostgreSQL to 18.7, 17.12, 16.16, 15.20, or 14.25 for the matching major line, then follow PGDG's post-update reindex and incremental-backup steps. 14.25 is the final PostgreSQL 14 release; plan your upgrade path off 14."
What to do about it
- Update now: Upgrade PostgreSQL to 18.7, 17.12, 16.16, 15.20, or 14.25 for the matching major line, then follow PGDG's post-update reindex and incremental-backup steps. 14.25 is the final PostgreSQL 14 release; plan your upgrade path off 14.
- Who should act: IT teams, managed service providers, and developers who run or ship type confusion ALTER SET DATA TYPE. Check every place it runs, including servers, build systems, containers, and older machines nobody watches closely.
- Read the official advisory from PostgreSQL (linked under Sources) for exact fixed builds and any temporary workaround if you cannot patch today.
- After patching, confirm the new version is actually running and review logs from November 12, 2026 onward for anything unusual.
- Home users: you do not need to buy or download anything because of CVE-2026-6460. A caller, text, or pop-up claiming your computer is "infected" with this CVE and asking for money or remote access is a scam.
CVE-2026-6460 record details from CISA and NVD
- Attack complexity for CVE-2026-6460: low
- CVE-2026-6460 CVSS scope: unchanged
- NVD sub-scores for CVE-2026-6460: exploitability 2.8, impact 5.9.
- Who mapped the weakness type: f86ef6dc-4d3a-42ad-8f28-e6d5547a5007 (Secondary)
- References in the NVD record: 1.
Dates to know
- November 12, 2026: NVD publishes CVE-2026-6460.
- November 12, 2026: FraudMeNot checks the CISA KEV catalog; CVE-2026-6460 is not listed.
- November 12, 2026: most recent NVD update (record status: Received).
Official references
Links listed in the NVD record (labels in brackets come from NVD):
Sources
General education only, not security advice. Facts checked against the sources above on the date shown.