The Persistent Vulnerability of Aligned AI Systems

• URL: http://arxiv.org/abs/2604.00324v1
• Date: Tue, 31 Mar 2026 23:49:07 GMT
• Title: The Persistent Vulnerability of Aligned AI Systems
• Abstract summary: This thesis contributes to four open problems in AI safety.<n> understanding dangerous internal computations, removing dangerous behaviors once embedded, testing for vulnerabilities before deployment, and predicting when models will act against deployers.
• Score: 2.9994384641583185
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Autonomous AI agents are being deployed with filesystem access, email control, and multi-step planning. This thesis contributes to four open problems in AI safety: understanding dangerous internal computations, removing dangerous behaviors once embedded, testing for vulnerabilities before deployment, and predicting when models will act against deployers. ACDC automates circuit discovery in transformers, recovering all five component types from prior manual work on GPT-2 Small by selecting 68 edges from 32,000 candidates in hours rather than months. Latent Adversarial Training (LAT) removes dangerous behaviors by optimizing perturbations in the residual stream to elicit failure modes, then training under those perturbations. LAT solved the sleeper agent problem where standard safety training failed, matching existing defenses with 700x fewer GPU hours. Best-of-N jailbreaking achieves 89% attack success on GPT-4o and 78% on Claude 3.5 Sonnet through random input augmentations. Attack success follows power law scaling across text, vision, and audio, enabling quantitative forecasting of adversarial robustness. Agentic misalignment tests whether frontier models autonomously choose harmful actions given ordinary goals. Across 16 models, agents engaged in blackmail (96% for Claude Opus 4), espionage, and actions causing death. Misbehavior rates rose from 6.5% to 55.1% when models stated scenarios were real rather than evaluations. The thesis does not fully resolve any of these problems but makes each tractable and measurable.

Related papers

• Cybersecurity Detection Classification with Reasoning-enabled Language Models [50.72675435043546]
  A major issue in Security Operations Centers (SOCs) is alert fatigue.<n>Prior work prompts or fine-tunes large language models (LLMs) to emit a triage label directly, but does not train them to reason about whether a detection is a genuine threat.<n>We train a chain-of-thought (CoT) reasoning-enabled triage classifier on real, human-labeled Windows endpoint detections.
  arXiv  Detail & Related papers  (2026-07-30T16:22:13Z)
• Understanding and Evaluating Claw-like Agent Security Through a Computer-Systems Lens [65.53494487819053]
  Claw-like AI agents (e.g., OpenClaw) are always-on processes with persistent access to credentials, files, tools, and external services.<n>We adopt a computer-system analogy: treating a Claw-like agent as an agentic computer system whose gateway runtime plays an OS-like mediation role.<n>We develop SafeClawArena, a benchmark of 406 adversarial tasks across four attack surfaces.
  arXiv  Detail & Related papers  (2026-06-29T18:00:45Z)
• Poster: ClawdGo: Endogenous Security Awareness Training for Autonomous AI Agents [14.881952206331968]
  ClawdGo is a framework for endogenous security awareness training.<n>Agents deployed on platforms such as OpenClaw face prompt injection, memory poisoning, supply-chain attacks, and social engineering.<n>We teach the agent to recognise and reason about threats from the inside, at inference time, with no model modification.
  arXiv  Detail & Related papers  (2026-04-27T04:09:27Z)
• The Blind Spot of Agent Safety: How Benign User Instructions Expose Critical Vulnerabilities in Computer-Use Agents [29.98056047452529]
  Computer-use agents (CUAs) can now autonomously complete complex tasks in real environments, but when misled, they can also be automated.<n>We introduce OSBLIND, a benchmark that evaluates CUAs under unintended attack conditions.<n>Most CUAs exceed 90% attack success rate (ASR), and even safety-aligned Claude 4.5 Sonnet reaches 72.7% ASR.
  arXiv  Detail & Related papers  (2026-04-12T10:52:42Z)
• Your Agent, Their Asset: A Real-World Safety Analysis of OpenClaw [87.97230960702274]
  We present the first real-world safety evaluation of OpenClaw.<n>We introduce the CIK taxonomy, which unifies an agent's persistent state into three dimensions.<n>Our evaluations cover 12 attack scenarios on a live OpenClaw instance.
  arXiv  Detail & Related papers  (2026-04-06T15:27:05Z)
• SafetyDrift: Predicting When AI Agents Cross the Line Before They Actually Do [0.0]
  SafetyDrift models agent safety trajectories as absorbing Markov chains.<n>In communication tasks, agents that reach even a mild risk state have an 85% chance of violating safety within five steps.<n>A monitor built on these models detects 94.7% of violations with 3.7 steps of advance warning at negligible computational cost.
  arXiv  Detail & Related papers  (2026-03-28T05:52:04Z)
• How Vulnerable Are AI Agents to Indirect Prompt Injections? Insights from a Large-Scale Public Competition [48.32744727426218]
  LLM based agents are increasingly deployed in high stakes settings where they process external data sources such as emails, documents, and code repositories.<n>This creates exposure to indirect prompt injection attacks, where adversarial instructions embedded in external content manipulate agent behavior without user awareness.<n>We present findings from a large scale public red teaming competition evaluating this dual objective across three agent settings.
  arXiv  Detail & Related papers  (2026-03-16T14:49:36Z)
• Internal Safety Collapse in Frontier Large Language Models [65.00730294617382]
  This work identifies a critical failure mode in frontier large language models (LLMs)<n>Under certain task conditions, models enter a state in which they continuously generate harmful content while executing otherwise benign tasks.<n>We introduce TVD, a framework that triggers ISC through domain tasks where generating harmful content is the only valid completion.
  arXiv  Detail & Related papers  (2026-03-04T12:55:34Z)
• Penetration Testing of Agentic AI: A Comparative Security Analysis Across Models and Frameworks [0.0]
  Agentic AI introduces security vulnerabilities that traditional LLM safeguards fail to address.<n>We conduct the first systematic testing and comparative evaluation of agentic AI systems.<n>We identify six distinct defensive behavior patterns including a novel "hallucinated compliance" strategy.
  arXiv  Detail & Related papers  (2025-12-16T19:22:50Z)
• Echoes of Human Malice in Agents: Benchmarking LLMs for Multi-Turn Online Harassment Attacks [10.7231991032233]
  Large Language Model (LLM) agents are powering a growing share of interactive web applications, yet remain vulnerable to misuse and harm.<n>We present the Online Harassment Agentic Benchmark consisting of: (i) a synthetic multi-turn harassment conversation dataset, (ii) a multi-agent (e.g., harasser, victim) simulation informed by repeated game theory, (iii) three jailbreak methods attacking agents across memory, planning, and fine-tuning, and (iv) a mixed-methods evaluation framework.
  arXiv  Detail & Related papers  (2025-10-16T01:27:44Z)
• Measuring Harmfulness of Computer-Using Agents [14.509501604704127]
  Computer-using agents (CUAs) can autonomously control computers to perform multi-step actions.<n> CUAHarm consists of 104 expert-written realistic misuse risks.<n>We evaluate frontier LMs including GPT-5, Claude 4 Sonnet, Gemini 2.5 Pro, Llama-3.3-70B, and Mistral Large 2.
  arXiv  Detail & Related papers  (2025-07-31T07:02:19Z)
• AegisLLM: Scaling Agentic Systems for Self-Reflective Defense in LLM Security [74.22452069013289]
  AegisLLM is a cooperative multi-agent defense against adversarial attacks and information leakage.<n>We show that scaling agentic reasoning system at test-time substantially enhances robustness without compromising model utility.<n> Comprehensive evaluations across key threat scenarios, including unlearning and jailbreaking, demonstrate the effectiveness of AegisLLM.
  arXiv  Detail & Related papers  (2025-04-29T17:36:05Z)
• Safety Pretraining: Toward the Next Generation of Safe AI [68.99129474671282]
  We present a data-centric pretraining framework that builds safety into the model from the start.<n>Our framework consists of four key steps: Safety Filtering, Safety Rephrasing, Native Refusal and Harmfulness-Tag annotated pretraining.<n>Our safety-pretrained models reduce attack success rates from 38.8% to 8.4% on standard LLM safety benchmarks with no performance on general degradation tasks.
  arXiv  Detail & Related papers  (2025-04-23T17:58:08Z)
• IDEATOR: Jailbreaking and Benchmarking Large Vision-Language Models Using Themselves [70.43466586161345]
  IDEATOR is a novel jailbreak method that autonomously generates malicious image-text pairs for black-box jailbreak attacks.<n>Our benchmark results on 11 recently releasedVLMs reveal significant gaps in safety alignment.<n>For instance, our challenge set ASRs of 46.31% on GPT-4o and 19.65% on Claude-3.5-Sonnet, underscoring the urgent need for stronger defenses.
  arXiv  Detail & Related papers  (2024-10-29T07:15:56Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.