security audit teams

Abstract: Binary patch presence testing (PPT) determines whether a security fix is present in a stripped binary. Prior systems often assume that patch-relevant code is known or successfully matched. Function-level matching is brittle to compilation changes, inlining, finite model context, and missing usage context; moreover, similarity suppresses the patch-sized distinction needed for the verdict. We present PatchAlign, an end-to-end binary PPT workflow, which performs large-scale localization that maximizes recall economically, then the final verdict reasons for patch-specific evidence. PatchAlign uses instruction-aligned retrieval and caller-callee context to accurately locate patch-relevant binary code pieces, then screens the shortlist and compares target binary code with vulnerable and patched references using an LLM. Our experiments indicate that PatchAlign can precisely localize patch-relevant code and determine patch status without a target-location oracle.

Abstract: As language-model-based coding agents evolve their programs and reuse code from one another, safety failures can acquire a history that extends beyond the agent in which they are detected. An unsafe change may originate in one agent, propagate through intermediate recipients, and survive subsequent revisions before an audit reveals it. Understanding this history is important for identifying the introducing change and distinguishing inherited behavior from independently occurring failures. We study origin attribution in self-evolving multi-agent coding systems: recovering the agent and program version that introduced an observed unsafe behavior into the target's code lineage, or determining that it arose locally. This task is challenging because rewriting obscures textual similarity, independent occurrences confound temporal ordering, and intermediate recipients can resemble the source. We propose a method that combines behavioral fingerprint matching with historical localization and recursive tracing. It compares candidate implementations beyond the unsafe outcome, retains evidence from multiple historical versions, and follows matching candidates upstream. On 158 queries across 85 controlled scenarios, the method achieves 96.8% exact attribution accuracy, compared with 77.2% for exhaustive fingerprint matching without recursion. Component ablations and evaluation on 34 additional queries using the same method and hyperparameters demonstrate the complementary roles of multiple representatives and recursive tracing. Further checks extend the evaluation to approval-boundary and resource-owner errors. Together, these results show how behavioral and historical evidence can recover the origins of unsafe behavior through code revision and cross-agent propagation.

Abstract: Security evaluation of learning-based systems requires more than just testing the system against a fixed collection of attacks. It requires adaptive mechanisms that can efficiently discover \textit{sets} of inputs that induce model failure. We introduce ATLAS (Adaptive Trust-Regions for Latent Adversarial Searches), which is a query-based framework that discovers adversarial input sets for black-box learning systems. ATLAS casts attack generation as an active learning level set estimation problem then combines calibrated approximations with a local-global sampling architecture to find regions of the input space that contain adversarial examples. Once discovered, ATLAS is designed to sample points within these adversarial regions to build adversarial sets that accurately represent the state of robustness of the target model. When applied on toy experiments, we find that ATLAS is able to recover more of the adversarial region under a limited query budget than does previous work. When applied to standard and adversarially trained MNIST, CIFAR, and ImageNet model targets, ATLAS produces better representative attacks than other query-based black-box attacks (NES, SignHunter, BayesOpt). ATLAS represents an automated red-teaming framework that can be used for both analyzing the robustness of learning-based systems under development and continuous auditing to see how the robustness of a system changes over time.

Abstract: Coding agents are now proficient enough to generate complex software applications from a single prompt. As their capabilities have grown, human oversight has increasingly shifted from line-by-line code review toward hands-off evaluation of outcomes. However, recent studies have shown that such a transition exposes a critical risk: functional correctness alone does not guarantee a secure implementation. Despite growing attention to code security, training safer coding agents remains challenging because reliable security supervision is difficult to obtain at scale from real-world repositories. We introduce AuraForge to synthesize and validate executable security tests for training secure coding agents. Our approach combines attack-oriented test synthesis, language-extensible task construction, and safeguards against reward hacking. Using AuraForge, we construct AuraGym, a multi-language and multi-CWE executable training gym: 679 executable feature-implementation tasks from 344 real-world repositories across Python, JavaScript, and TypeScript, covering 177 CWE categories. On the subset with human-written security tests, AuraForge produces about 3 times as many test cases on average and reduces the false-positive rate by 83.23%, allowing alternative secure implementations to receive correct supervision. Training Qwen3.5-4B with synthesized security tests gains larger improvements than human-written security tests (average 19.7 FuncPass and 6.2 SecPass vs. 14.9 FuncPass and 4.4 SecPass) on three languages. These results demonstrate that AuraForge provides more diverse and reliable security supervision to train secure coding agents.

Abstract: Natural language autoencoders translate a language model's internal activations into readable explanations. Explaining every token position is costly. Which positions should an auditor inspect to understand a potential threat? We study this question across $4.7$ million explanations on prompt injection and concealment. We compare signals from model computation with a ranker trained only on chat structure. Chat structure usually selects more relevant explanations than the computational signals, without requiring a model forward pass for position selection. On three of four datasets, explaining just $5\%$ of positions retains nearly all of the success rate from explaining every position, where success means obtaining an explanation about the threat. The benefit varies with the audit task. We also show that pretrained verbalizers recover words that models have learned to conceal through fine-tuning, without additional verbalizer training. These results identify where auditors can concentrate explanation generation and show that useful explanations can extend beyond the model a verbalizer was trained to describe.

Abstract: AI-powered development tools are now widely used to generate code and assist developers with routine programming tasks. Although existing work has identified vulnerabilities in AI-generated code, security-oriented work is often focused on vulnerability detection rather than risk assessment. To address this gap, this paper presents a Security Risk Assessment Framework (SRF) to evaluate the security risks of AI-generated code. SRF combines threat modeling, security analysis, and a quantitative risk evaluation approach based on vulnerability criticality. The framework is applied to a set of security-relevant programming tasks, where code generated by multiple AI-powered development tools is analyzed using Bandit and Semgrep. The results show that AI-generated code can introduce security vulnerabilities across all evaluated tools. They also show that risk levels vary by task type, as input processing and file handling tasks showed higher risk, while simpler tasks remained low-risk. Differences between tools exist but are smaller than differences across task categories. Overall, SRF enables reproducible evaluation of AI-generated code and provides a practical framework for assessing its security implications.

Abstract: Cross-chain bridges move value between blockchains. Intent-based bridges are a variant where a solver fulfills a user's declared outcome and an off-chain settlement layer later reconciles the fill against the deposit. Existing smart-contract fuzzers and static analyzers only flag known-bad code patterns or require protocol-specific hand-written assertions. This work formalizes a taxonomy separating invariant violations, safety properties a contract must enforce locally, from settlement exposures legitimately delegated to the off-chain settlement layer, and proposes IntentFuzz: a protocol-aware fuzzer that recovers a bridge's intent structure and deposit/fill function roles directly from unannotated Solidity source, then synthesizes multi-step fuzz sequences using an LLM-based fallback to help build call arguments. IntentFuzz recovers the correct intent structure in 9/9 benchmark protocols and classifies deposit and fill functions with 100% recall and 82% combined precision; across a corpus of 77 manually labeled contracts, it reaches 79.5% bridge-classification precision and 97.2% recall, and among confirmed bridges, struct selection reaches 88.6% precision and recall while deposit and fill classification each reach 100% recall. On 23 planted-bug mutants, IntentFuzz attains 100% recall and 100% precision, executing 273 templates (507 transactions in a median of 14ms per template). Across 24 real-world deployments, it confirms 17 genuine invariant violations under heuristic-only input generation, rising to 22 with its LLM-assisted tier enabled, spanning eight vulnerable GitHub repositories, each finding reproducible against public, deployed bytecode.

Abstract: Conventional vulnerability analysis relies on either system access or dynamic interaction, all of which may be unavailable to third-party analysts auditing closed-source, remotely hosted, critical in situ systems, or commercially gated software. Therefore, we propose a new paradigm of no-box vulnerability analysis in which neither access nor runtime interaction is available, and only functionality metadata is available. Such metadata defines the intended behavior of the system, including its inputs, outputs, and side effects, while constraining the space of implementations consistent with that behavior. We propose hypothesizing about vulnerabilities that exist across all possible implementations of a given system metadata, without observing or interacting with the target system. An analyst can later validate these hypotheses when additional access is available. We showcase the feasibility of no-box vulnerability analysis through implementing a prototype called MCPSEC, which audits Model Context Protocol (MCP) servers for indirect prompt injection vulnerabilities using only the tool metadata exposed at server registration time. We evaluate MCPSEC on 20 widely deployed MCP servers comprising 177 tools, among which human evaluators confirm 95 vulnerable tools. MCPSEC identified 143 tools as vulnerable, and for each vulnerable tool, it produced a hypothesized vulnerability along with exploitation technique. Using metadata alone, MCPSEC predicted 94 (98.9% recall) real verified vulnerabilities, compared against an LLM baseline with 80 (84.2% recall). Overall, our results introduce no-box vulnerability analysis as a new analysis paradigm and demonstrate its practical feasibility in realistic systems.