CVE-2026-59319: CWE-943 Improper Neutralization of Special Elements in Data Query Logic in Spring Spring AI

The vulnerability arises because RedisChatMemoryRepository.findByMetadata() builds RediSearch tag and text queries from caller-supplied metadata without applying the necessary escaping function RediSearchUtil.escape(). This contrasts with other methods like get(), clear(), and findByTimeRange() which properly escape inputs. An attacker who can supply user-controlled values to findByMetadata() on a tag-typed metadata field can inject RediSearch syntax (e.g., x} | *) that breaks out of the tag clause, resulting in queries that match all indexed chat messages across every conversation in the index. The affected product version is Spring AI 2.0.0.

The vulnerability allows an attacker with the ability to supply user-controlled metadata values to craft queries that bypass intended tag restrictions, potentially exposing all indexed chat messages across conversations. The CVSS score of 4.3 reflects a low confidentiality impact with no integrity or availability impact. There is no indication of known exploits in the wild.

No official patch or remediation is currently available for this vulnerability. Users should monitor the vendor advisory for updates. Until a fix is released, avoid passing untrusted user input directly to the findByMetadata() method on tag-typed metadata fields or implement manual escaping of special characters in such inputs.