标题： AI代理智能合约漏洞生成 显示英文标题

标题： AI Agent Smart Contract Exploit Generation

摘要： 我们提出A1，一个由代理执行驱动的系统，可将任何大语言模型转化为端到端的漏洞利用生成器。 A1没有手工设计的启发式方法，并为代理提供了六个特定领域的工具，使其能够自主发现漏洞。 代理可以灵活地利用这些工具来理解智能合约行为，生成利用策略，在区块链状态上进行测试，并根据执行反馈优化方法。 所有输出都经过具体验证，以消除误报。 在以太坊和币安智能链上的36个现实世界易受攻击的合约评估中，A1在VERITE基准测试中的成功率为62.96%（27个中有17个）。 除了VERITE数据集外，A1还发现了9个额外的易受攻击的合约，其中5个案例发生在最强模型训练截止日期之后。 在所有26个成功案例中，A1每个案例提取高达859万美元，总计933万美元。 通过在六个大语言模型上的432次实验，我们分析了迭代性能，显示平均边际收益分别为+9.7%、+3.7%、+5.1%和+2.8%，分别对应迭代2-5，每次实验的成本范围为$0.01-$3.59。 对19次历史攻击的蒙特卡洛分析显示，在没有检测延迟的情况下，成功率在85.9%-88.8%之间。 我们研究了攻击者或防御者在部署A1作为持续链上扫描系统时，谁最受益。 我们的模型显示，OpenAI的o3-pro在0.100%的漏洞发生率下，最多可承受30天的扫描延迟仍保持盈利，而更快的模型需要>=1.000%的比率才能收支平衡。 研究结果揭示了一个令人担忧的不对称性：在0.1%的漏洞率下，攻击者在6000美元的漏洞价值下实现链上扫描盈利，而防御者则需要60000美元，这引发了关于AI代理是否必然偏向于利用而非防御的根本性问题。 显示英文摘要

摘要： We present A1, an agentic execution driven system that transforms any LLM into an end-to-end exploit generator. A1 has no hand-crafted heuristics and provides the agent with six domain-specific tools that enable autonomous vulnerability discovery. The agent can flexibly leverage these tools to understand smart contract behavior, generate exploit strategies, test them on blockchain states, and refine approaches based on execution feedback. All outputs are concretely validated to eliminate false positives. The evaluation across 36 real-world vulnerable contracts on Ethereum and Binance Smart Chain demonstrates a 62.96% (17 out of 27) success rate on the VERITE benchmark. Beyond the VERITE dataset, A1 identified 9 additional vulnerable contracts, with 5 cases occurring after the strongest model's training cutoff date. Across all 26 successful cases, A1 extracts up to 8.59 million USD per case and 9.33 million USD total. Through 432 experiments across six LLMs, we analyze iteration-wise performance showing diminishing returns with average marginal gains of +9.7%, +3.7%, +5.1%, and +2.8% for iterations 2-5 respectively, with per-experiment costs ranging $0.01-$3.59. A Monte Carlo analysis of 19 historical attacks shows success probabilities of 85.9%-88.8% without detection delays. We investigate whether an attacker or a defender benefits most from deploying A1 as a continuous on-chain scanning system. Our model shows that OpenAI's o3-pro maintains profitability up to a 30.0 days scanning delay at 0.100% vulnerability incidence rates, while faster models require >=1.000% rates to break-even. The findings exposes a troubling asymmetry: at 0.1% vulnerability rates, attackers achieve an on-chain scanning profitability at a \$6000 exploit value, while defenders require \$60000, raising fundamental questions about whether AI agents inevitably favor exploitation over defense.