PostgreSQL pgcrypto CVE-2026-14663

Exploitation requires that the pgcrypto extension is installed and actively used in the target PostgreSQL instance, AND that at least one OpenSSL cipher used by the affected PGP functions has been explicitly disabled in the runtime OpenSSL configuration or is absent from the installed OpenSSL version - this is the critical, non-default prerequisite and the primary limiting factor. Environments running standard OpenSSL with common cipher suites enabled are not vulnerable. The confidentiality impact (cleartext recovery) additionally requires the attacker to gain access to the stored or transmitted data that was processed by the faulty cipher path. The integrity impact (wrong-key decryption success, MDC bypass enabling data injection) further requires the application to accept and process attacker-supplied PGP-encrypted blobs as input. PostgreSQL instances not using pgcrypto, and those whose OpenSSL configuration leaves the relevant ciphers enabled, are entirely unaffected regardless of PostgreSQL version.

The official CVSS 3.1 score of 6.5 with vector AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N warrants critical examination on the AC metric. The AC:L (low complexity) rating overstates ease of exploitation: the vulnerability manifests only when at least one OpenSSL cipher used by pgcrypto is explicitly disabled in the runtime OpenSSL configuration or absent from the installed OpenSSL build - a non-default, environment-specific condition. In practice, standard OpenSSL deployments with typical cipher suites enabled would not trigger the vulnerable code path, suggesting AC:H is more accurate for real-world assessments. Notably, organizations that have hardened their OpenSSL cipher configuration for compliance reasons (e.g., disabling older or export-grade ciphers) represent a realistic at-risk population, which is an ironic outcome of applying security hardening. The C:L and I:L impact ratings are well-calibrated: partial confidentiality loss for cipher-disabled data and modest integrity impact via the MDC bypass enabling potential data injection are realistic. EPSS data is not provided in the available intelligence, so quantitative exploitation probability cannot be assessed. KEV status is negative - no active exploitation is confirmed by CISA. No POC exploit code has been identified. Overall, real-world risk is lower than the raw CVSS score implies due to the prerequisite OpenSSL cipher configuration, but environments with intentional cipher restrictions face an unexpected and significant cryptographic failure.

An attacker with read access to database storage or application output containing pgcrypto-encrypted PGP data - in an environment where OpenSSL has disabled the relevant cipher - finds the stored ciphertext is actually cleartext, directly recoverable without any cryptographic key. Alternatively, against an application that accepts and decrypts PGP-encrypted input blobs, an attacker could craft a malformed PGP blob using the disabled cipher; the application's call to a pgcrypto decrypt function would return attacker-controlled plaintext with any supplied key, bypassing MDC integrity verification and potentially injecting malicious data into the application's processing pipeline. No public exploit code has been identified at the time of analysis.