ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?

ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?

AI agents are rapidly gaining capabilities that could significantly reshape cybersecurity, making rigorous evaluation urgent. A critical capability is exploitation: turning a vulnerability, which is not yet an attack, into a concrete security impact, such as unauthorized file access or code execution. Exploitation is a particularly challenging task because it requires low-level program reasoning (e.g., about memory layout), runtime adaptation, and sustained progress over long horizons. Meanwhile, it is inherently dual-use, supporting defensive workflows while lowering the barrier for offense. Despite its importance and diagnostic value, exploitation remains under-evaluated. To address this gap, we introduce ExploitGym, a large-scale, diverse, realistic benchmark on the exploitation capabilities of AI agents. Given a program input that triggers a vulnerability, ExploitGym tasks agents with progressively extending it into a working exploit. The benchmark comprises 898 instances sourced from real-world vulnerabilities across three domains, including userspace programs, Google's V8 JavaScript engine, and the Linux kernel. We vary the security protections applied to each instance, isolating their impact on agent performance. All configurations are packaged in reproducible containerized environments. Our evaluation shows that while exploitation remains challenging, frontier models can successfully exploit a non-trivial fraction of vulnerabilities. For example, the strongest configurations are Anthropic's latest model Claude Mythos Preview and OpenAI's GPT-5.5, which produce working exploits for 157 and 120 instances, respectively. Notably, even with widely used defenses enabled, models retain non-trivial success rates. These results establish ExploitGym as an effective testbed for exploitation and highlight the growing cybersecurity risks posed by increasingly capable AI agents.

CyberGym: Evaluating AI Agents' Real-World Cybersecurity Capabilities at Scale

AI agents have significant potential to reshape cybersecurity, making a thorough assessment of their capabilities critical. However, existing evaluations fall short, because they are based on small-scale benchmarks and only measure static outcomes, failing to capture the full, dynamic range of real-world security challenges. To address these limitations, we introduce CyberGym, a large-scale benchmark featuring 1,507 real-world vulnerabilities across 188 software projects. Adjustable to different vulnerability analysis settings, CyberGym primarily tasks agents with generating a proof-of-concept test that reproduces a vulnerability, given only its text description and the corresponding codebase. Our extensive evaluation highlights that CyberGym effectively differentiates agents' and models' cybersecurity capabilities. Even the top-performing combinations only achieve a ~20% success rate, demonstrating the overall difficulty of CyberGym. Beyond static benchmarking, we show that CyberGym leads to the discovery of 34 zero-day vulnerabilities and 18 historically incomplete patches. These results underscore that CyberGym is not only a robust benchmark for measuring AI's progress in cybersecurity but also a platform for creating direct, real-world security impact.

AI Agent Traps

As autonomous AI agents increasingly navigate the web, they face a novel challenge: the information environment itself. This gives rise to a critical vulnerability we refer to as "AI Agent Traps", i.e. adversarial content designed to manipulate, deceive, or exploit visiting agents. In this paper, we introduce the first known systematic framework for understanding this emerging threat. We break down how these traps work, identifying six types of attack: Content Injection Traps that exploit the gap between human perception, machine parsing, and dynamic rendering; Semantic Manipulation Traps, which corrupt an agent's reasoning and internal verification processes; Cognitive State Traps, which target an agent's long-term memory, knowledge bases, and learned behavioural policies; Behavioural Control Traps, which hijack an agent's capabilities to force unauthorised actions; Systemic Traps, which use agent interaction to create systemic failure, and Human-in-the-Loop Traps, which exploit cognitive biases to influence a human overseer. This research is not specific to any particular agent or model. By mapping this new attack surface, we identify critical gaps in current defences and propose a research agenda that could secure the entire agent ecosystem.

sunblaze-ucb/cybergym

CyberGym is a large-scale, high-quality cybersecurity evaluation framework designed to rigorously assess the capabilities of AI agents on real-world vulnerability analysis tasks.

Quantifying Frontier LLM Capabilities for Container Sandbox Escape

AI agents are demonstrating rapid improvement in capabilities: the length of some tasks that frontier models can complete autonomously—measured in human-equivalent time—has been doubling approximately every seven months (METR, 2025; AI Security Institute, 2025a). In cybersecurity, current models achieve non-trivial success (Zhang et al., 2025) on professional-level Capture the Flag challenges, and recent evaluations report 13% success rates on exploiting real-world web application vulnerabilities (Zhu et al., 2025b). These results indicate that modern models can already perform multi-step vulnerability discovery and exploitation.

Defeating Prompt Injections by Design

Large Language Models (LLMs) are increasingly deployed in agentic systems that interact with an untrusted environment. However, LLM agents are vulnerable to prompt injection attacks when handling untrusted data. In this paper we propose CaMeL, a robust defense that creates a protective system layer around the LLM, securing it even when underlying models are susceptible to attacks. To operate, CaMeL explicitly extracts the control and data flows from the (trusted) query; therefore, the untrusted data retrieved by the LLM can never impact the program flow. To further improve security, CaMeL uses a notion of a capability to prevent the exfiltration of private data over unauthorized data flows by enforcing security policies when tools are called. We demonstrate effectiveness of CaMeL by solving $77\%$ of tasks with provable security (compared to $84\%$ with an undefended system) in AgentDojo. We release CaMeL at https://github.com/google-research/camel-prompt-injection.

AgentSpec: Customizable Runtime Enforcement for Safe and Reliable LLM Agents

Agents built on LLMs are increasingly deployed across diverse domains, automating complex decision-making and task execution. However, their autonomy introduces safety risks, including security vulnerabilities, legal violations, and unintended harmful actions. Existing mitigation methods, such as model-based safeguards and early enforcement strategies, fall short in robustness, interpretability, and adaptability. To address these challenges, we propose AgentSpec, a lightweight domain-specific language for specifying and enforcing runtime constraints on LLM agents. With AgentSpec, users define structured rules that incorporate triggers, predicates, and enforcement mechanisms, ensuring agents operate within predefined safety boundaries. We implement AgentSpec across multiple domains, including code execution, embodied agents, and autonomous driving, demonstrating its adaptability and effectiveness. Our evaluation shows that AgentSpec successfully prevents unsafe executions in over 90% of code agent cases, eliminates all hazardous actions in embodied agent tasks, and enforces 100% compliance by autonomous vehicles (AVs). Despite its strong safety guarantees, AgentSpec remains computationally lightweight, with overheads in milliseconds. By combining interpretability, modularity, and efficiency, AgentSpec provides a practical and scalable solution for enforcing LLM agent safety across diverse applications. We also automate the generation of rules using LLMs and assess their effectiveness. Our evaluation shows that the rules generated by OpenAI o1 achieve a precision of 95.56% and recall of 70.96% for embodied agents, successfully identify 87.26% of the risky code, and prevent AVs from breaking laws in 5 out of 8 scenarios.

Microsoft Build 2026: Securing code, agents, and models across the development lifecycle | Microsoft Security Blog

Discover how Microsoft enables fast, secure AI development with MDASH and new security capabilities.

Top AI Security Incidents of 2025 Revealed | Adversa AI

Discover how AI systems are being hacked in the wild — from prompt injection to agent abuse — with real breaches, lessons, and defenses in Adversa AI’s 2025 report.

AI CVE Slop: The Crisis Drowning Open Source Security

The proliferation of AI-generated vulnerability reports — commonly termed “AI slop” — has emerged as one of the most significant…

On the Coming Industrialisation of Exploit Generation with LLMs

Recently I ran an experiment where I built agents on top of Opus 4.5 and GPT-5.2 and then challenged them to write exploits for a zeroday vulnerability in the QuickJS Javascript interpreter. I adde…

SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?

Large Language Model (LLM) agents have been widely adopted in modern software development workflows. SWE-bench [13] and related works [23, 24, 22, 25, 15] establish the task of issue resolution as a de-facto standard for assessing their capability and usefulness. In this setting, an agent is given an entire codebase, a task description (e.g., a bug report or feature request) in natural language and is instructed to produce a code patch that resolves the issue and passes the repository’s test suite. These benchmarks have been instrumental in demonstrating both the substantial potential and the persistent limitations of current models as SWE agents.

SCAM — How safe is your AI agent?

An open-source benchmark by 1Password that tests whether AI agents can handle real security threats during everyday tasks.