CVE-2026-108955 Tiiny-AI CVE debrief

Who should care

Defenders of systems using Tiiny-AI PowerInfer, especially those with local access and untrusted input, should assess exposure and prioritize verification. This includes operators, platform administrators, and security teams responsible for vulnerability management and incident response. They should review CVE and NVD details, assess local access controls, and consider compensating controls for exposed systems. Monitoring for potential exploitation and re-

Why it matters

CVE-2026-108955 is a medium-severity vulnerability in Tiiny-AI PowerInfer requiring local access for exploitation. Defenders should assess exposure, prioritize verification, and consider compensating controls.

• Verify input validation for gguf_init_from_file function
• Assess local access controls for ggml.c
• Monitor for potential exploitation attempts
• Consider compensating controls for untrusted input

Technical summary

The gguf_init_from_file function in ggml.c of Tiiny-AI PowerInfer is vulnerable to an out-of-bounds read. This requires local access to exploit. The product's rolling release model means specific affected or updated version information is not available. Defenders should focus on verifying input validation and assessing local access controls. Compensating controls may be necessary for systems with untrusted input. Monitoring for exploitation attempts is also recommended. The CVE record provides limited detail, so defenders must consider potential impacts based on similar vulnerabilities.

Defensive priority

Assess exposure and prioritize verification for local systems using Tiiny-AI PowerInfer, especially those with untrusted input.

Recommended defensive actions

• Inventory and assess systems using Tiiny-AI PowerInfer for potential exposure
• Verify local access controls and validate input to the gguf_init_from_file function
• Monitor for and address potential exploitation attempts
• Consider compensating controls for systems with untrusted input
• Review vendor guidance for updates or patches
• Assess vulnerability management processes for Tiiny-AI PowerInfer
• Track changes to Tiiny-AI PowerInfer for future updates

Evidence notes

The vulnerability is in the gguf_init_from_file function of ggml.c in Tiiny-AI PowerInfer. Local access is required for exploitation. The product uses a rolling release model, making specific version information for affected or updated releases unavailable. Evidence is limited to CVE and NVD details. Defenders should verify input validation for gguf_init_from_file function and assess local access controls for ggml.c.

Sources and references

Methodology and review provenance

AI-assisted synthesis based on stored public vulnerability evidence. System validation, approval state, and publication status do not by themselves establish human review of this revision. PatchSiren helps prioritize defensive review and does not prove exposure or remediation on any system.