Cybersecurity AI agents invalidate the CRA's premises about vulnerability discovery, exploitation, and remediation speed, making static conformity certificates false and requiring continuous agent-operated defense.
hub Theoretical anchor
the winning worker cost
Theoretical anchor. 60% of citing Pith papers extend or build on this work's framework.
20
Pith papers citing it
19
external citations · Pith
Extended
60% of classified citations
abstract
LLM agents have become increasingly sophisticated, especially in the realm of cybersecurity. Researchers have shown that LLM agents can exploit real-world vulnerabilities when given a description of the vulnerability and toy capture-the-flag problems. However, these agents still perform poorly on real-world vulnerabilities that are unknown to the agent ahead of time (zero-day vulnerabilities). In this work, we show that teams of LLM agents can exploit real-world, zero-day vulnerabilities. Prior agents struggle with exploring many different vulnerabilities and long-range planning when used alone. To resolve this, we introduce HPTSA, a system of agents with a planning agent that can launch subagents. The planning agent explores the system and determines which subagents to call, resolving long-term planning issues when trying different vulnerabilities. We construct a benchmark of 14 real-world vulnerabilities and show that our team of agents improve over prior agent frameworks by up to 4.3X.
hub tools
citation-role summary
background 5
citation-polarity summary
roles
background 5representative citing papers
ZERO-APT is a closed-loop framework that integrates an LLM attacker, configurable LLM defender, and judge agent to achieve 79% attack success rate, 0.860 causal consistency, and full decision auditability in penetration testing under intelligent defense.
Empirical study of 400 LLM attack runs finds exploitation success rates of 25-85% across four models against a fixed multi-service honeypot, with model-distinctive failure modes and p<0.001 differences.
SETC framework provides the first systematic comparison of CIM, OCSF, and ECS logging standards by running 50 RCE exploits and measuring how well each captures attack indicators.
Trace fingerprints AI penetration testing agents from terminal command sequences to identify model families and extracts their system prompts via targeted defensive prompt injection.
Dynamic Cyber Ranges with LLM defender agents reduce attacker success to 0-55% and preserve evaluation headroom as models advance by using comparable capabilities on both sides.
A new 7x4 taxonomy organizes agentic AI security threats by architectural layer and persistence timescale, revealing under-explored upper layers and missing defenses after surveying 116 papers.
LLMVD.js uses LLM agents to confirm 84% of taint-style vulnerabilities on public benchmarks (vs. <22% for prior tools) and generates validated exploits for 36 of 260 new packages (vs. ≤2 for traditional tools).
LLM agents outperform humans in romance-baiting scams, eliciting greater trust and 46% compliance versus 18%, with 0% detection by safety filters and 87% of scam tasks automatable.
Frontier AI agents frequently violate corrigibility by overriding interruptions in benign computer-use tasks, with misalignment increasing alongside model capability.
An agentic pipeline localizes the security-relevant function in 10 of 20 Ubuntu binary security updates and produces an accepted root-cause classification in 11 of 20, limited mainly by binary differencing coverage.
Empirical comparison of agentic topologies for offensive security shows MAS-Indep reaching 64.2% validated detection while simpler baselines remain competitive on efficiency, with whitebox and web targets outperforming blackbox and binary ones.
RSA prompting enables LLMs to automatically create functional exploits for CVEs in Odoo ERP, succeeding on all tested cases in 3-5 rounds and removing the need for manual effort.
FORGE deploys a fixed five-agent pipeline on 603 CVEs to achieve 67.8% L1+ exploitation success at $1.50 per CVE while generating detection rules whose grounding improves with deeper exploitation traces.
Vulnsage, a multi-agent framework, generates 34.64% more exploits than prior tools and verified 146 zero-day vulnerabilities in real-world open-source libraries.
xOffense automates penetration testing via a fine-tuned Qwen3-32B LLM in a multi-agent setup with specialized agents for reconnaissance, vulnerability scanning, and exploitation, reporting 79.17% sub-task completion on AutoPenBench and AI-Pentest-Benchmark.
A systematic review of neuro-symbolic AI in cybersecurity finds that deeper integration and causal reasoning improve performance across intrusion detection and vulnerability tasks, while identifying barriers and a research roadmap.
The paper proposes the Cybersecurity AI Scientist as a modular multi-agent architecture for automating cybersecurity research, distinguished by its focus on non-stationary threats and anchored in a four-zeros risk-trust-incident-energy frame.
A benchmarking experiment finds low rediscovery rates for three models on six Mythos-linked bug tasks, with only six target matches across 54 attempts under controlled prompting.
Agentic AI lowers the cost and speed of cyber attacks, requiring immediate improvements in identity management, phishing-resistant authentication, patching, and agent governance for large enterprises and the Mittelstand.
citing papers explorer
-
Certifying Ghosts: How Cybersecurity AI Agents Break the EU Cyber Resilience Act
Cybersecurity AI agents invalidate the CRA's premises about vulnerability discovery, exploitation, and remediation speed, making static conformity certificates false and requiring continuous agent-operated defense.
-
ZERO-APT: A Closed-Loop Adversarial Framework for LLM-Driven Automated Penetration Testing under Intelligent Defense
ZERO-APT is a closed-loop framework that integrates an LLM attacker, configurable LLM defender, and judge agent to achieve 79% attack success rate, 0.860 causal consistency, and full decision auditability in penetration testing under intelligent defense.
-
How Reliable Are AI Attackers Against a Fixed Vulnerable Target? A 400-Run Empirical Study of LLM Penetration Testing Consistency
Empirical study of 400 LLM attack runs finds exploitation success rates of 25-85% across four models against a fixed multi-service honeypot, with model-distinctive failure modes and p<0.001 differences.
-
Beyond Collection: Measuring the Detection Efficacy of Modern Security Logging Standards
SETC framework provides the first systematic comparison of CIM, OCSF, and ECS logging standards by running 50 RCE exploits and measuring how well each captures attack indicators.
-
Trace: Unmasking AI Attack Agents Through Terminal Behavior Fingerprinting
Trace fingerprints AI penetration testing agents from terminal command sequences to identify model families and extracts their system prompts via targeted defensive prompt injection.
-
Dynamic Cyber Ranges
Dynamic Cyber Ranges with LLM defender agents reduce attacker success to 0-55% and preserve evaluation headroom as models advance by using comparable capabilities on both sides.
-
A Systematic Survey of Security Threats and Defenses in LLM-Based AI Agents: A Layered Attack Surface Framework
A new 7x4 taxonomy organizes agentic AI security threats by architectural layer and persistence timescale, revealing under-explored upper layers and missing defenses after surveying 116 papers.
-
Taint-Style Vulnerability Detection and Confirmation for Node.js Packages Using LLM Agent Reasoning
LLMVD.js uses LLM agents to confirm 84% of taint-style vulnerabilities on public benchmarks (vs. <22% for prior tools) and generates validated exploits for 36 of 260 new packages (vs. ≤2 for traditional tools).
-
Love, Lies, and Language Models: Investigating AI's Role in Romance-Baiting Scams
LLM agents outperform humans in romance-baiting scams, eliciting greater trust and 46% compliance versus 18%, with 0% detection by safety filters and 87% of scam tasks automatable.
-
ROGUE: Misaligned Agent Behavior Arising from Ordinary Computer Use
Frontier AI agents frequently violate corrigibility by overriding interruptions in benign computer-use tasks, with misalignment increasing alongside model capability.
-
Patch2Vuln: Agentic Reconstruction of Vulnerabilities from Linux Distribution Binary Patches
An agentic pipeline localizes the security-relevant function in 10 of 20 Ubuntu binary security updates and produces an accepted root-cause classification in 11 of 20, limited mainly by binary differencing coverage.
-
Towards Optimal Agentic Architectures for Offensive Security Tasks
Empirical comparison of agentic topologies for offensive security shows MAS-Indep reaching 64.2% validated detection while simpler baselines remain competitive on efficiency, with whitebox and web targets outperforming blackbox and binary ones.
-
From Rookie to Expert: Manipulating LLMs for Automated Vulnerability Exploitation in Enterprise Software
RSA prompting enables LLMs to automatically create functional exploits for CVEs in Odoo ERP, succeeding on all tested cases in 3-5 rounds and removing the need for manual effort.
-
FORGE: Multi-Agent Graduated Exploitation and Detection Engineering
FORGE deploys a fixed five-agent pipeline on 603 CVEs to achieve 67.8% L1+ exploitation success at $1.50 per CVE while generating detection rules whose grounding improves with deeper exploitation traces.
-
A Multi-Agent Framework for Automated Exploit Generation with Constraint-Guided Comprehension and Reflection
Vulnsage, a multi-agent framework, generates 34.64% more exploits than prior tools and verified 146 zero-day vulnerabilities in real-world open-source libraries.
-
xOffense: An Autonomous Multi-Agent Framework for Penetration Testing with Domain-Adapted Large Language Models
xOffense automates penetration testing via a fine-tuned Qwen3-32B LLM in a multi-agent setup with specialized agents for reconnaissance, vulnerability scanning, and exploitation, reporting 79.17% sub-task completion on AutoPenBench and AI-Pentest-Benchmark.
-
Neuro-Symbolic AI for Cybersecurity: State of the Art, Challenges, and Opportunities
A systematic review of neuro-symbolic AI in cybersecurity finds that deeper integration and causal reasoning improve performance across intrusion detection and vulnerability tasks, while identifying barriers and a research roadmap.
-
Hephaestus: Toward a Cybersecurity AI Scientist
The paper proposes the Cybersecurity AI Scientist as a modular multi-agent architecture for automating cybersecurity research, distinguished by its focus on non-stationary threats and anchored in a four-zeros risk-trust-incident-energy frame.
-
Benchmarking Mythos-Linked Bug Rediscovery
A benchmarking experiment finds low rediscovery rates for three models on six Mythos-linked bug tasks, with only six target matches across 54 attempts under controlled prompting.
-
Agentic AI and the Industrialization of Cyber Offense: Forecast, Consequences, and Defensive Priorities for Enterprises and the Mittelstand
Agentic AI lowers the cost and speed of cyber attacks, requiring immediate improvements in identity management, phishing-resistant authentication, patching, and agent governance for large enterprises and the Mittelstand.