CVE-2026-39909: Use After Free in ggml-org llama.cpp
A use-after-free vulnerability exists in llama.cpp before commit b8585 in the RPC server's GRAPH_RECOMPUTE handler. This flaw allows unauthenticated remote attackers to achieve arbitrary read and write access by manipulating stored computation graphs and triggering re-execution with dangling pointers. Exploitation can lead to full remote code execution without authentication or user interaction.
AI Analysis
Technical Summary
CVE-2026-39909 is a critical use-after-free vulnerability in the llama.cpp project (ggml-org) affecting versions before commit b8585. The vulnerability resides in the RPC server's GRAPH_RECOMPUTE handler, where an attacker can store a computation graph, free referenced buffers, and then reclaim freed memory with attacker-controlled content. By sending crafted RPC requests, the attacker triggers re-execution of stored graphs containing dangling pointers, enabling arbitrary read/write access and remote code execution without requiring authentication or user interaction. The CVSS 4.0 score is 9.2, reflecting the critical impact and network attack vector with no privileges or user interaction needed.
Potential Impact
Successful exploitation allows unauthenticated remote attackers to execute arbitrary code remotely on the affected system by exploiting use-after-free conditions in the RPC server. This can lead to full system compromise without any user interaction or authentication.
Mitigation Recommendations
Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. No official fix or patch links are provided at this time. Until a patch is available, restrict network access to the RPC server to trusted users only and monitor for suspicious RPC requests targeting GRAPH_RECOMPUTE functionality.
CVE-2026-39909: Use After Free in ggml-org llama.cpp
Description
A use-after-free vulnerability exists in llama.cpp before commit b8585 in the RPC server's GRAPH_RECOMPUTE handler. This flaw allows unauthenticated remote attackers to achieve arbitrary read and write access by manipulating stored computation graphs and triggering re-execution with dangling pointers. Exploitation can lead to full remote code execution without authentication or user interaction.
CVSS v4.0
Score 9.2critical
Affected software
Run on your own infrastructure? Check whether these packages are installed with threat-finder — our free open-source scanner.
AI-Powered Analysis
Machine-generated threat intelligence
Technical Analysis
CVE-2026-39909 is a critical use-after-free vulnerability in the llama.cpp project (ggml-org) affecting versions before commit b8585. The vulnerability resides in the RPC server's GRAPH_RECOMPUTE handler, where an attacker can store a computation graph, free referenced buffers, and then reclaim freed memory with attacker-controlled content. By sending crafted RPC requests, the attacker triggers re-execution of stored graphs containing dangling pointers, enabling arbitrary read/write access and remote code execution without requiring authentication or user interaction. The CVSS 4.0 score is 9.2, reflecting the critical impact and network attack vector with no privileges or user interaction needed.
Potential Impact
Successful exploitation allows unauthenticated remote attackers to execute arbitrary code remotely on the affected system by exploiting use-after-free conditions in the RPC server. This can lead to full system compromise without any user interaction or authentication.
Mitigation Recommendations
Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. No official fix or patch links are provided at this time. Until a patch is available, restrict network access to the RPC server to trusted users only and monitor for suspicious RPC requests targeting GRAPH_RECOMPUTE functionality.
Technical Details
- Data Version
- 5.2
- Assigner Short Name
- VulnCheck
- Date Reserved
- 2026-04-07T20:57:06.209Z
- Cvss Version
- 4.0
- State
- PUBLISHED
- Remediation Level
- null
Threat ID: 6a888260acd9273b496b4aaa
Added to database: 08/21/2026, 16:52:48 UTC
Last enriched: 08/21/2026, 17:07:07 UTC
Last updated: 08/21/2026, 19:29:33 UTC
Views: 7
Community Reviews
0 reviewsCrowdsource mitigation strategies, share intel context, and vote on the most helpful responses. Sign in to add your voice and help keep defenders ahead.
Want to contribute mitigation steps or threat intel context? Sign in or create an account to join the community discussion.
Actions
Updates to AI analysis require Pro Console access. Upgrade inside Console → Billing.
Need more coverage?
Upgrade to Pro Console for AI refresh and higher limits.
For incident response and remediation, OffSeq services can help resolve threats faster.
Latest Threats
Check if your credentials are on the dark web
Instant breach scanning across billions of leaked records. Free tier available.