arXiv subjects

Byoungyoung Lee

Publications and source records attributed to Byoungyoung Lee.

7 recordsLinked to original sources

arXiv2026

DualView: Preventing Indirect Prompt Injection in Personal AI Agents

Personal AI agents that run on the user's local machine automate daily tasks including web search, email, and file management. Their access to computer resources, including the network, file system, and shell, exposes them to indirect prompt injection (IPI) attacks. Prior Dual LLM defenses block IPI by replacing untrusted data with symbols that the agent can reference but not read. However, they track untrusted data only inside the agent's context, so when the agent saves and later rereads untrusted data, that data, possibly an attacker's prompt, can return as trusted data rather than as a symbol, which we call stored IPI. Operating on the user's real environment is what makes agents like OpenClaw practical, and is exactly why a defense that ignores it is incomplete. Preserving symbols in such an environment is hard, because humans and programs need original data. We present DualView, which extends untrusted data tracking from the agent's context to the user's environment, including the file system, shell, network, and other agents, by giving each channel two views. In AgentView, the agent sees untrusted data as symbols even after writing it out and reading it back, blocking stored IPI, while HumanView preserves original data for humans and tools. DualView routes each tool call to the right view and synchronizes data across the two views. DualView deploys as an OpenClaw plugin using only tool hooks, without changing the agent's tool-call logic or tool implementations. DualView deterministically prevents instructions in untrusted data from directly steering the agent's tool calls; this guarantee does not depend on recognizing the evaluated attack templates. In our evaluation on an IPI benchmark and PinchBench, DualView blocked every tested IPI attack, including stored IPI. On PinchBench, its utility drop was within 1.8 to 6.4 points.

cs.CR↗

arXiv2026

Agent Data Injection Attacks are Realistic Threats to AI Agents

AI agents act on behalf of user prompts, consuming external data and taking actions based on the agent context. Prior research on AI agent security has primarily focused on indirect prompt injection (IPI). Its most well-studied category is instruction injection, where attacker-controlled untrusted data is interpreted as an instruction. In response, many mitigations have been proposed to prevent instruction injection attacks. In this paper, we introduce a new category of IPI, agent data injection attacks (ADI). ADI injects malicious data disguised as trusted data, such as security-critical metadata (e.g., resource identifiers or data origins) or agent context data (e.g., tool call and response formats). As a result, agents unknowingly execute unintended actions based on attacker-controlled data. ADI has similar attack impacts as instruction injection attacks, because it causes agents to misbehave and execute unintended actions. Despite the similar impact, ADI remains underexplored and easily bypasses existing IPI defenses. We found several critical vulnerabilities in real-world agents that allow an attacker to launch various attacks: arbitrary click attacks on web agents (Claude in Chrome, Antigravity, and Nanobrowser), and remote code execution and supply-chain attacks on coding agents (Claude Code, Codex, and Gemini CLI). We evaluate ADI vulnerabilities across off-the-shelf models and AI agents, and find that ADI is effective in both standalone LLMs and AI agent settings. ADI exposes a critical gap in agent security, signifying that current AI agents do not employ a fundamental security principle: current agents do not isolate trusted data from untrusted data.

cs.CR↗

arXiv2025

Prompt Flow Integrity to Prevent Privilege Escalation in LLM Agents

Large Language Models (LLMs) are combined with tools to create powerful LLM agents that provide a wide range of services. Unlike traditional software, LLM agent's behavior is determined at runtime by natural language prompts from either user or tool's data. This flexibility enables a new computing paradigm with unlimited capabilities and programmability, but also introduces new security risks, vulnerable to privilege escalation attacks. Moreover, user prompts are prone to be interpreted in an insecure way by LLM agents, creating non-deterministic behaviors that can be exploited by attackers. To address these security risks, we propose Prompt Flow Integrity (PFI), a system security-oriented solution to prevent privilege escalation in LLM agents. Analyzing the architectural characteristics of LLM agents, PFI features three mitigation techniques -- i.e., agent isolation, secure untrusted data processing, and privilege escalation guardrails. Our evaluation result shows that PFI effectively mitigates privilege escalation attacks while successfully preserving the utility of LLM agents.

cs.CR↗

arXiv2024

TEEMATE: Fast and Efficient Confidential Container using Shared Enclave

Confidential container is becoming increasingly popular as it meets both needs for efficient resource management by cloud providers, and data protection by cloud users. Specifically, confidential containers integrate the container and the enclave, aiming to inherit the design-wise advantages of both (i.e., resource management and data protection). However, current confidential containers suffer from large performance overheads caused by i) a larger startup latency due to the enclave creation, and ii) a larger memory footprint due to the non-shareable characteristics of enclave memory. This paper explores a design conundrum of confidential container, examining why the confidential containers impose such large performance overheads. Surprisingly, we found there is a universal misconception that an enclave can only be used by a single (containerized) process that created it. However, an enclave can be shared across multiple processes, because an enclave is merely a set of physical resources while the process is an abstraction constructed by the host kernel. To this end, we introduce TeeMate, a new approach to utilize the enclaves on the host system. Especially, TeeMate designs the primitives to i) share the enclave memory between processes, thus preserving memory abstraction, and ii) assign the threads in enclave between processes, thus preserving thread abstraction. We concretized TeeMate on Intel SGX, and implemented confidential serverless computing and confidential database on top of TeeMate based confidential containers. The evaluation clearly demonstrated the strong practical impact of TeeMate by achieving at least 4.5 times lower latency and 2.8 times lower memory usage compared to the applications built on the conventional confidential containers.

cs.CR↗

arXiv2024

TikTag: Breaking ARM's Memory Tagging Extension with Speculative Execution

ARM Memory Tagging Extension (MTE) is a new hardware feature introduced in ARMv8.5-A architecture, aiming to detect memory corruption vulnerabilities. The low overhead of MTE makes it an attractive solution to mitigate memory corruption attacks in modern software systems and is considered the most promising path forward for improving C/C++ software security. This paper explores the potential security risks posed by speculative execution attacks against MTE. Specifically, this paper identifies new TikTag gadgets capable of leaking the MTE tags from arbitrary memory addresses through speculative execution. With TikTag gadgets, attackers can bypass the probabilistic defense of MTE, increasing the attack success rate by close to 100%. We demonstrate that TikTag gadgets can be used to bypass MTE-based mitigations in real-world systems, Google Chrome and the Linux kernel. Experimental results show that TikTag gadgets can successfully leak an MTE tag with a success rate higher than 95% in less than 4 seconds. We further propose new defense mechanisms to mitigate the security risks posed by TikTag gadgets.

cs.CR↗

arXiv2019

A Tale of Two Trees: One Writes, and Other Reads. Optimized Oblivious Accesses to Large-Scale Blockchains

The Bitcoin network has offered a new way of securely performing financial transactions over the insecure network. Nevertheless, this ability comes with the cost of storing a large (distributed) ledger, which has become unsuitable for personal devices of any kind. Although the simplified payment verification (SPV) clients can address this storage issue, a Bitcoin SPV client has to rely on other Bitcoin nodes to obtain its transaction history and the current approaches offer no privacy guarantees to the SPV clients. This work presents $T^3$, a trusted hardware-secured Bitcoin full client that supports efficient oblivious search/update for Bitcoin SPV clients without sacrificing the privacy of the clients. In this design, we leverage the trusted execution and attestation capabilities of a trusted execution environment (TEE) and the ability to hide access patterns of oblivious random access memory (ORAM) to protect SPV clients' requests from a potentially malicious server. The key novelty of $T^3$ lies in the optimizations introduced to conventional ORAM, tailored for expected SPV client usages. In particular, by making a natural assumption about the access patterns of SPV clients, we are able to propose a two-tree ORAM construction that overcomes the concurrency limitation associated with traditional ORAMs. We have implemented and tested our system using the current Bitcoin Unspent Transaction Output database. Our experiment shows that the system is feasible to be deployed in practice while providing strong privacy and security guarantees to Bitcoin SPV clients.

cs.CR↗

arXiv2019

SynFuzz: Efficient Concolic Execution via Branch Condition Synthesis

Concolic execution is a powerful program analysis technique for exploring execution paths in a systematic manner. Compare to random-mutation-based fuzzing, concolic execution is especially good at exploring paths that are guarded by complex and tight branch predicates (e.g., (a*b) == 0xdeadbeef). The drawback, however, is that concolic execution engines are much slower than native execution. One major source of the slowness is that concolic execution engines have to the interpret instructions to maintain the symbolic expression of program variables. In this work, we propose SynFuzz, a novel approach to perform scalable concolic execution. SynFuzz achieves this goal by replacing interpretation with dynamic taint analysis and program synthesis. In particular, to flip a conditional branch, SynFuzz first uses operation-aware taint analysis to record a partial expression (i.e., a sketch) of its branch predicate. Then it uses oracle-guided program synthesis to reconstruct the symbolic expression based on input-output pairs. The last step is the same as traditional concolic execution - SynFuzz consults a SMT solver to generate an input that can flip the target branch. By doing so, SynFuzz can achieve an execution speed that is close to fuzzing while retain concolic execution's capability of flipping complex branch predicates. We have implemented a prototype of SynFuzz and evaluated it with three sets of programs: real-world applications, the LAVA-M benchmark, and the Google Fuzzer Test Suite (FTS). The evaluation results showed that SynFuzz was much more scalable than traditional concolic execution engines, was able to find more bugs in LAVA-M than most state-of-the-art concolic execution engine (QSYM), and achieved better code coverage on real-world applications and FTS.

cs.CR↗