From Attack Simulation to SIEM Rule: Deterministic Detection-as-Code Synthesis with Probe-Level Traceability

Rule-based Intrusion Detection and Prevention Systems (IDPS) offer precise attack detection as well as mitigation, however their manually crafted, signature-driven rules limit adaptability to emerging and zero-day threats. Additionally, existing public datasets (e.g., CICIDS2017, UNSW-NB15) focus on traffic classification and provide little structured information to support automatic rule synthesis or prevention logic. To address this gap, we propose Generative Thread Intelligence (GenTI) \footnote{GenTI refers to the proposed framework, and GTI refers to the dataset.} an LLM-driven benchmark for automatic generation of IDPS rules targeting unseen attacks. The dataset (GTI) aggregates over 150k detection and prevention rules from Snort, Suricata, Emerging Threats, as well as 50k YARA, each annotated with protocol behavior, payload signatures, contextual relationships, mappings to Cyber Threat Intelligence (CTI), along with actionable response types (alert, drop, reject). Moreover, on top of this corpus we design an LLM-based pipeline that transforms analyst prompts and representative payloads into deployable rules via structured prompt engineering, Chain-of-Thought (CoT) reasoning, as well as a Chain-of-Verification (CoVe) loop for syntactic, semantic, and security validation. The generated rules are executed in real time on (Snort/Suricata) and evaluated by syntax accuracy, semantic similarity, CTI coverage, security effectiveness as well as unseen attacks detection. Furthermore, our GenTI instantiation achieves a composite rule-quality score of 89.4%, with 94.8% CTI coverage, improving unseen attacks detection from 45% to 87.4% and reducing the false-positive rate from 8.5% to 2.3%. Overall, GenTI establishes the first large-scale benchmark that tightly couples rule-level CTI with LLM-based automation, enabling adaptive, self-evolving IDPS.

Security Information and Event Management (SIEM) systems aggregate log data from heterogeneous sources to detect coordinated attacks. Traditional rule-based correlation engines struggle to classify multi-step web application attacks because they examine each event without reference to the behavioural history of the originating host. We present Smart-SIEM, an AI module for the open-source Wazuh SIEM platform with two contributions: (1) a per-source-IP behavioural context vector encoding HTTP response-status distributions, peak rule activation counts, and MITRE ATT&CK technique frequencies from the N most recent prior events; (2) a two-stage hybrid cascade combining LightGBM for binary attack detection and XGBoost for six-class attack categorisation. Evaluated on 46,454 purpose-built Wazuh security events, context features improve all tested gradient boosting algorithms from ~0.705 macro F1 to 0.947-0.967 (Stage 1) and 0.876-0.914 (Stage 2), an average gain of +0.254 and +0.324 respectively. The hybrid cascade achieves F1 of 0.967 (binary) and 0.914 (six-class). Wazuh's native rule engine detects 0% of Brute Force and Broken Authentication events; the AI module detects 100% and 98.3% respectively. A self-adaptive retraining mechanism recovers from concept drift: F1 drops from 0.905 to 0.465 when unseen attack types emerge, recovering to 0.814 after retraining on the combined corpus.

Production LLM applications stack several defense families -- refusal-phrase filters, token-budget controls, model allowlists, rate limits, tool-registry authentication -- yet existing breach-and-attack-simulation (BAS) benchmarks report a single aggregate coverage number, hiding which family closes which threat. We measure attribution. We add four OWASP-LLM-Top-10-aware agents to a 21-agent baseline scanner and target a lattice of four synthetic LLM endpoints: $L_0$ (no defenses), $L_1$ (refusal-only), $L_2$ (budget-only), and $L_3$ (full stack). $L_1$ and $L_2$ are sibling single-axis ablations, not subsets of each other; $L_3$ is their union plus tool-registry authentication and credential scrubbing. Across $N=10$ replications, the per-OWASP finding count is clean: refusal alone removes all LLM01 (jailbreak) and LLM07 (system-prompt leakage) findings; budget alone removes all LLM02 (sensitive-info disclosure) and LLM10 (unbounded consumption) findings by terminating multi-step sequences; LLM06 (excessive agency) requires the full stack. We probe brittleness under paraphrasing: with 300 Gemini-generated paraphrases ($K=5$ over a 60-template brittleness corpus), $L_1$ refusal block rate falls 15 pp on LLM01 and 25 pp on LLM07. A fifth target, $L_4$-real, swaps the stub backend for Gemini-2.5-flash behind the same $L_3$ regex and matches $L_1$ exactly, indicating no measurable alignment contribution beyond the regex (not a general claim about alignment). Budget controls show no drop (0 pp once the rate-limit floor is factored out). A refusal whitelist that clears a static benchmark can be defeated by an LLM-driven paraphraser without changing attack intent; a budget control resists the same mutation.