AI Security Research Portal
Sources

Open Weight AI SOC Paper Collection

Untrusted source collection. Use individual source records for claims.

Scope

This collection tracks papers and thesis/report-style sources where open-weight or open-source models are used, evaluated, audited, or proposed for SOC/security-operations workflows.

Newly Collected Sources

SourceKindOpen-weight relevanceRaw file
When the Ruler is Broken: Parsing-Induced Suppression in LLM-Based Security Log EvaluationpreprintAudits OpenSOC-AI/TinyLlama evaluation and proposes SOC-Bench v0.when-ruler-broken-parsing-induced-suppression-soc-log-eval-2026
Llama-3.1-FoundationAI-SecurityLLM-Base-8B Technical ReportpreprintOpen-weight cybersecurity foundation model built on Llama 3.1.foundation-sec-8b-base-technical-report-2025
Llama-3.1-FoundationAI-SecurityLLM-8B-Instruct Technical ReportpreprintInstruction-tuned open-weight cybersecurity assistant model for professional workflows.foundation-sec-8b-instruct-technical-report-2025
Evaluation of LLM Agents for the SOC Tier 1 Analyst Triage ProcessthesisSOC Tier 1 triage workflow includes Llama 3 70B model use.soc-tier1-llm-agents-triage-thesis-2024

Existing Related Sources Already In Vault

SourceWhy relevant
opensoc-ai-parameter-efficient-log-analysis-2026TinyLlama-1.1B + LoRA for threat classification, MITRE ATT&CK mapping, and severity assessment from raw logs.
soc-ai-companion-wosoc-2026Locally deployed SOC assistant paper; current source should be rechecked for gpt-oss/open-weight deployment details.
journal-big-data-llm-soc-risk-management-2026Compares GPT-3.5 Turbo and Mistral-7B on SOC-oriented network telemetry tasks.
siabench-security-incident-analysis-2026Evaluates open-weight and closed-weight LLMs for security incident analysis with SIA Agent/SIABENCH.
e-mantra-cognitive-threat-detection-soc-2026WOSOC 2026 SOC-oriented LLM routing paper; mentions open-source alternatives such as Llama-3 70B for API dependency mitigation.

Recommended Ingest Questions