Executive brief
OpenSIPS is an open-source SIP server used by telecommunications providers to handle voice and video traffic. A security flaw in its JWT authentication module allows attackers to bypass security checks by sending a specially crafted token. This could allow an unauthorized user to impersonate legitimate identities and gain access to restricted communication services.
Technical details
A SQL injection vulnerability exists in the jwt_db_authorize() function within modules/auth_jwt/authorize.c of OpenSIPS. When db_mode is enabled, the application extracts the 'tag' claim from a JSON Web Token (JWT) without prior signature verification and interpolates the unescaped value directly into a SQL query. An attacker can exploit this by crafting a JWT with a malicious SQL payload in the tag claim (e.g., using a UNION SELECT statement) to inject a known secret into the query result. Because OpenSIPS uses the result of this query to verify the JWT's signature, the attacker can then sign the token with their own secret to successfully bypass authentication. This vulnerability is addressed in OpenSIPS version 3.6.4 and commit 3822d33.
Affected products
- OpenSIPS OpenSIPS 3.1 before 3.6.4
Timeline
- 2026-02-02: patched: Fix merged into master branch
- 2026-02-18: advisory: OpenSIPS 3.6.4 released with fix
- 2026-02-25: disclosed: NVD publication date
References
- https://github.com/OpenSIPS/opensips/commit/3822d33c1c6b25832fdd88da1d23eed74be55b05
- https://github.com/OpenSIPS/opensips/pull/3807
- https://opensips.org/
- https://opensips.org/pub/opensips/3.6.4/ChangeLog
- https://www.vulncheck.com/advisories/opensips-auth-jwt-sql-injection-enables-jwt-authentication-bypass