Synergistic Directed Execution and LLM-Driven Analysis for Zero-Day AI-Generated Malware Detection

Dimensions overall score 9.0

GitHub Code Pulse

No public code linked for this paper yet.

• The paper introduces a hybrid analysis framework combining concolic execution with LLM-augmented path prioritization and deep-learning-based vulnerability classification for zero-day AI-generated malware detection.method 95%
  Evidencepartial

  This paper introduces a novel hybrid analysis framework that synergistically combines \emph{concolic execution} with \emph{LLM-augmented path prioritization} and \emph{deep-learning-based vulnerability classification} to detect zero-day AI-generated malware with provable guarantees.

  Implicationmissing

  Implication not extracted yet.

  Verificationpartial

  partial

• The LLM-guided concolic exploration strategy reduces the average number of explored paths by 73.2% compared to depth-first search while maintaining equivalent malicious-path coverage.result 90%
  Evidencepartial

  The framework introduces three novel algorithms: (i) an LLM-guided concolic exploration strategy that reduces the average number of explored paths by 73.2\% compared to depth-first search while maintaining equivalent malicious-path coverage

  Implicationmissing

  Implication not extracted yet.

  Verificationpartial

  partial

• The framework achieves 98.7% accuracy on conventional malware and 97.5% accuracy on AI-generated threats.result 90%
  Evidencepartial

  Experimental evaluation on the EMBER, Malimg, SOREL-20M, and a novel AI-Gen-Malware benchmark comprising 2{,}500 LLM-synthesized samples demonstrates that achieves 98.7\% accuracy on conventional malware and 97.5\% accuracy on AI-generated threats

  Implicationmissing

  Implication not extracted yet.

  Verificationpartial

  partial

• The framework outperforms ClamAV, YARA, MalConv, and EMBER-GBDT baselines by margins of 8.4-52.2 percentage points on AI-generated samples.result 90%
  Evidencepartial

  outperforming ClamAV, YARA, MalConv, and EMBER-GBDT baselines by margins of 8.4--52.2 percentage points on AI-generated samples.

  Implicationmissing

  Implication not extracted yet.

  Verificationpartial

  partial

• The paper proves both soundness and relative completeness of the detection algorithm, assuming classifier correctness.technical 85%
  Evidencepartial

  and prove both the \emph{soundness} and \emph{relative completeness} of our detection algorithm, assuming classifier correctness.

  Implicationmissing

  Implication not extracted yet.

  Verificationpartial

  partial

• The framework includes a transformer-based path-constraint classifier trained on symbolic execution traces.method 95%
  Evidencepartial

  (ii) a transformer-based path-constraint classifier trained on symbolic execution traces

  Implicationmissing

  Implication not extracted yet.

  Verificationpartial

  partial

• The framework includes a feedback loop that iteratively refines the LLM's prioritization policy using reinforcement learning from detection outcomes.method 95%
  Evidencepartial

  (iii) a feedback loop that iteratively refines the LLM's prioritization policy using reinforcement learning from detection outcomes.

  Implicationmissing

  Implication not extracted yet.

  Verificationpartial

  partial

• AI-generated malware exhibits polymorphic, metamorphic, and context-aware evasion capabilities that render signature-based and shallow heuristic defenses obsolete.technical 85%
  Evidencepartial

  AI-generated malware exhibits polymorphic, metamorphic, and context-aware evasion capabilities that render signature-based and shallow heuristic defenses obsolete.

  Implicationmissing

  Implication not extracted yet.

  Verificationpartial

  partial

• The framework achieves 98.7% accuracy on conventional malware and 97.5% accuracy on AI-generated threats.result 95%
  Evidencepartial

  Experimental evaluation ... demonstrates that achieves 98.7% accuracy on conventional malware and 97.5% accuracy on AI-generated threats

  Implicationpartial

  Directly stated in abstract with specific numeric values.

  Verificationpartial

  partial

• The proposed hybrid analysis framework achieves 98.7% accuracy on conventional malware and 97.5% accuracy on AI-generated threats.result 95%
  Evidencepartial

  Experimental evaluation on the EMBER, Malimg, SOREL-20M, and a novel AI-Gen-Malware benchmark comprising 2,500 LLM-synthesized samples demonstrates that achieves 98.7% accuracy on conventional malware and 97.5% accuracy on AI-generated threats

  Implicationpartial

  Directly stated in the abstract with specific numeric values.

  Verificationpartial

  partial

• The framework outperforms ClamAV, YARA, MalConv, and EMBER-GBDT baselines by margins of 8.4–52.2 percentage points on AI-generated samples.result 95%
  Evidencepartial

  outperforming ClamAV, YARA, MalConv, and EMBER-GBDT baselines by margins of 8.4–52.2 percentage points on AI-generated samples

  Implicationpartial

  Directly stated in the abstract with specific numeric range.

  Verificationpartial

  partial

• The LLM-guided concolic exploration strategy reduces the average number of explored paths by 73.2% compared to depth-first search while maintaining equivalent malicious-path coverage.result 95%
  Evidencepartial

  an LLM-guided concolic exploration strategy that reduces the average number of explored paths by 73.2% compared to depth-first search while maintaining equivalent malicious-path coverage

  Implicationpartial

  Directly stated in the abstract with specific percentage and comparison.

  Verificationpartial

  partial

Startup potential card