Exploitability of malformed MobileIdentity5GS inputs accepted by the defective patch

Determine whether malformed pre-authentication NAS RegistrationRequest inputs containing violating MobileIdentity5GS values, which are silently accepted because the validation block in free5GC nas v1.2.3 is unreachable, are individually exploitable through downstream code.

Background

CVE-2025-69248 concerns an AMF panic in free5GC’s MobileIdentity5GS decoder caused by a malformed pre-authentication NAS RegistrationRequest. Although the v1.2.3 patch closes the original panic path, its additional defense-in-depth validation block is ineffective because it reads from a zero-allocated buffer before the wire bytes are copied and applies an incorrect mask to the identity-type field.

Seven unit tests demonstrate that malformed MobileIdentity5GS values—including undersized GUTI, IMEI, 5G-S-TMSI, and SUCI values, undefined identity types, and an oversized length—are accepted despite the intended validation. The paper explicitly leaves unresolved whether these accepted inputs are individually exploitable through downstream code.

References

The practical impact is the elimination of the intended defense-in-depth layer for a class of pre-authentication inputs whose individual exploitability remains an open question for downstream code.

Drishti: AI-Led Human-Directed Vulnerability Auditing for 5G Cores  (2608.30112 - Ramachandran et al., 31 Aug 2026) in Appendix, Section “CVE-2025-69248 patch defect: line-by-line analysis,” paragraph “Why no live-stack reproduction was needed” (Appendix B)