Research graph
References from A Survey on Autonomy-Induced Security Risks in Large Model-Based Agents. Local targets link to admitted publications; unresolved targets remain external evidence.
10.1016/j.inffus.2025.103599
10.1016/j.inffus.2025.103599 · External reference
10.1007/s11432-024-4222-0
10.1007/s11432-024-4222-0 · External reference
On the opportunities and risks of foundation models
2021 · External reference
GPT-4 Technical report
2023 · External reference
The LLaMA 3 herd of models
2024 · External reference
REACT: Synergizing reasoning and acting in language models
2023 · External reference
10.52202/075280-2997
10.52202/075280-2997 · External reference
WebGPT: Browser-assisted question-answering with human feedback
2021 · External reference
Voyager: An open-ended embodied agent with large language models
2024 · External reference
Auto-GPT for online decision making: Benchmarks and additional opinions
2023 · External reference
10.1109/raie66699.2025.00013
10.1109/raie66699.2025.00013 · External reference
AutoGen: Enabling next-gen LLM applications via multi-agent conversation
2024 · External reference
Unresolved reference
2021 · External reference
LLM powered autonomous agents
2023 · External reference
A survey on large language model based autonomous agents
10.1007/s11704-024-40231-1 · 2024 · External reference
Large language model agent: A survey on methodology, applications and challenges
2025 · External reference
Unresolved reference
2014 · External reference
10.1126/science.adn0117
10.1126/science.adn0117 · External reference
10.1007/978-3-642-32560-1_2
10.1007/978-3-642-32560-1_2 · External reference
10.1201/9781351251389-3
10.1201/9781351251389-3 · External reference
10.70777/si.v2i5.15569
10.70777/si.v2i5.15569 · External reference
International scientific report on the safety of advanced AI: Interim report
2024 · External reference
10.3389/fhumd.2025.1579166
10.3389/fhumd.2025.1579166 · External reference
10.1007/978-3-662-54033-6_5
10.1007/978-3-662-54033-6_5 · External reference
10.52202/075280-1480
10.52202/075280-1480 · External reference
AI agent security: Emerging threats and defense paradigms
2024 · External reference
AdvLM: Visual adversarial attacks on autonomous driving vision-language models
2025 · External reference
Agent Security Bench (ASB): Formalizing and benchmarking attacks and defenses in LLM-based agents
2025 · External reference
ASB: A security benchmark for LLM-based autonomous agents
2025 · External reference
10.18653/v1/2025.emnlp-main.1771
10.18653/v1/2025.emnlp-main.1771 · External reference
10.18653/v1/2024.findings-emnlp.585
10.18653/v1/2024.findings-emnlp.585 · External reference
10.1145/3711896.3736561
10.1145/3711896.3736561 · External reference
Commercial LLM agents are already vulnerable to simple yet dangerous attacks
2025 · External reference
The obvious invisible threat: LLM-powered GUI agents’ vulnerability to fine-print injections
2025 · External reference
10.1109/acai68217.2025.11406310
10.1109/acai68217.2025.11406310 · External reference
10.1002/widm.70111
10.1002/widm.70111 · External reference
MLATrust: Benchmarking trustworthiness of multimodal LLM agents in GUI environments
2025 · External reference
Alignment challenges in autonomous language agents
2024 · External reference
Caution for the environment: Multimodal agents are susceptible to environmental distractions
2024 · External reference
10.52202/079017-2636
10.52202/079017-2636 · External reference
Towards action hijacking of large language model-based agent
2024 · External reference
10.52202/079017-4136
10.52202/079017-4136 · External reference
10.18653/v1/2025.acl-long.413
10.18653/v1/2025.acl-long.413 · External reference
From exploration to mastery: Enabling LLMs to master tools via self-driven interactions
2025 · External reference
Technical risks of (lethal) autonomous weapons systems
2025 · External reference
Imprompter: Tricking LLM agents into improper tool use
2024 · External reference
Tool-induced misalignment in LLM agents: A causal analysis
2025 · External reference
Safeguarding AI agents: Developing and analyzing safety architectures
2024 · External reference
Levels of AGI for operationalizing progress on the path to AGI
2023 · External reference
10.70777/si.v2i3.15295
10.70777/si.v2i3.15295 · External reference
Levels of autonomy: Liability in the age of AI agents
2025 · External reference
10.18653/v1/2025.acl-long.1227
10.18653/v1/2025.acl-long.1227 · External reference
The landscape of emerging AI agent architectures for reasoning, planning, and tool calling: A survey
2024 · External reference
TPTU: Task planning and tool usage of large language model-based AI agents
2023 · External reference
10.1038/s42256-024-00832-8
10.1038/s42256-024-00832-8 · External reference
10.1109/iccv51701.2025.00323
10.1109/iccv51701.2025.00323 · External reference
10.18653/v1/2024.acl-long.292
10.18653/v1/2024.acl-long.292 · External reference
Agent Q: Advanced reasoning and learning for autonomous AI agents
2024 · External reference
Ignore previous prompt: Attack techniques for language models
2023 · External reference
Universal and transferable adversarial attacks on aligned language models
2023 · External reference
AutoGPT: Build, deploy, and run AI agents
2023 · External reference
10.52202/075280-0377
10.52202/075280-0377 · External reference
World models and active inference
2023 · External reference
Reinforcement learning with learned world models
2021 · External reference
AGI ruin: A list of lethalities
2022 · External reference
Deep reinforcement learning from human preferences
2017 · External reference
Refusal-trained LLMs are easily Jailbroken as browser agents
2024 · External reference
10.18653/v1/2024.findings-emnlp.79
10.18653/v1/2024.findings-emnlp.79 · External reference
The effect of state representation on LLM agent behavior in dynamic routing games
2025 · External reference
AgentAlign: Navigating safety alignment in the shift from informative to agentic large language models
2025 · External reference
Medical hallucinations in foundation models and their impact on healthcare
10.1101/2025.02.28.25323115 · 2025 · External reference
Fine-tuning language models from human preferences
2019 · External reference
Learning to summarize with human feedback
2020 · External reference
10.52202/068431-2011
10.52202/068431-2011 · External reference
Demonstrating specification gaming in reasoning models
2025 · External reference
10.1145/3773080
10.1145/3773080 · External reference
Concrete problems in AI safety
2016 · External reference
Empowering working memory for large language model agents
2023 · External reference
OpenAI red teaming network
2023 · External reference
A survey of red teaming in NLP: Attacks, evaluation, and defense
2022 · External reference
Ignore previous prompt: Attack techniques for language models
2022 · External reference
A trembling house of cards? Mapping adversarial attacks against language agents
2024 · External reference
10.18653/v1/2024.acl-long.530
10.18653/v1/2024.acl-long.530 · External reference
10.1109/tnnls.2018.2886017
10.1109/tnnls.2018.2886017 · External reference
AgentHarm: A benchmark for measuring harmfulness of LLM agents
2025 · External reference
10.52202/079017-1561
10.52202/079017-1561 · External reference
10.1109/comst.2023.3344808
10.1109/comst.2023.3344808 · External reference
SCAV: Steering LLM hallucination via self-checking augmented verification
2023 · External reference
Extracting training data from large language models
2021 · External reference
Membership inference attacks on sequence-to-sequence models
2022 · External reference
The surprising effectiveness of latent world models for continual reinforcement learning
2022 · External reference
Instruction tuning with GPT-4
2023 · External reference
Red teaming language models to reduce harms: Methods, scaling behaviors, and lessons learned
2022 · External reference
Retrieval-augmented language models expose sensitive information
2023 · External reference
Prompt injection attack against LLM-integrated applications
2023 · External reference
10.1145/3690624.3709179
10.1145/3690624.3709179 · External reference
10.18653/v1/2025.findings-acl.410
10.18653/v1/2025.findings-acl.410 · External reference
10.18653/v1/2025.findings-naacl.123
10.18653/v1/2025.findings-naacl.123 · External reference
Red teaming the mind of the machine: A systematic evaluation of prompt injection and jailbreak vulnerabilities in LLMs
2025 · External reference
How robust is google’s bard to adversarial image attacks?
2023 · External reference
Adversarial attacks on multimodal agents
2025 · External reference
Probing the robustness of vision–language pretrained models: A multimodal adversarial attack approach
2024 · External reference
10.52202/085713-0272
10.52202/085713-0272 · External reference
LLM01:2025 prompt injection
2025 · External reference
10.1609/aaai.v39i26.34947
10.1609/aaai.v39i26.34947 · External reference
10.52202/085713-1554
10.52202/085713-1554 · External reference
10.1609/aaaiss.v2i1.27688
10.1609/aaaiss.v2i1.27688 · External reference
10.18653/v1/2024.acl-long.830
10.18653/v1/2024.acl-long.830 · External reference
LLM as OS, agents as Apps: Envisioning AIOS, agents and the AIOS-agent ecosystem
2023 · External reference
Language models are few-shot learners
2020 · External reference
10.1201/9781351251389-16
10.1201/9781351251389-16 · External reference
Teams of LLM agents can exploit zero-day vulnerabilities
2024 · External reference
Cascading hallucination in multi-agent systems
2025 · External reference
Techniques and business perspectives LLM-based multi-agent systems
2024 · External reference
10.18653/v1/2025.findings-acl.349
10.18653/v1/2025.findings-acl.349 · External reference
10.1201/9781315140223
10.1201/9781315140223 · External reference
A survey of safe reinforcement learning and constrained MDPS: Foundations and extensions
2025 · External reference
Constrained policy optimization
2017 · External reference
Benchmarking safe exploration in deep reinforcement learning
2019 · External reference
10.52202/079017-4136
10.52202/079017-4136 · External reference
10.18653/v1/2025.findings-naacl.395
10.18653/v1/2025.findings-naacl.395 · External reference
10.1007/978-3-032-16092-8_28
10.1007/978-3-032-16092-8_28 · External reference
10.52202/079017-3201
10.52202/079017-3201 · External reference
10.18653/v1/2025.findings-naacl.363
10.18653/v1/2025.findings-naacl.363 · External reference
Multi-agent systems execute arbitrary malicious code
2025 · External reference
10.52202/085713-1440
10.52202/085713-1440 · External reference
The effects of reward misspecification: Mapping and mitigating misaligned models
2022 · External reference
10.52202/068431-0687
10.52202/068431-0687 · External reference
Reward hacking in reinforcement learning
2024 · External reference
MONA: Myopic optimization with non-myopic approval can mitigate multi-step reward hacking
2025 · External reference
10.1145/3716628
10.1145/3716628 · External reference
Debunc: Mitigating hallucinations in large language model agent communication with uncertainty estimations
2024 · External reference
10.2307/jj.1640535.12
10.2307/jj.1640535.12 · External reference
10.1016/s0019-9958(64)90223-2
10.1016/s0019-9958(64)90223-2 · External reference
Formalizing two problems of realistic world models
2015 · External reference
Vingean reflection: Reliable reasoning for self-improving agents
2026 · External reference
VILA: On pre-training for visual language models
2023 · External reference
10.24894/sprengglossarium_m03262
10.24894/sprengglossarium_m03262 · External reference
StruQ: Defending against prompt injection with structured queries
2024 · External reference
Context-aware drift detection
2022 · External reference
10.1109/ickg63256.2024.00067
10.1109/ickg63256.2024.00067 · External reference
10.52202/075280-0814
10.52202/075280-0814 · External reference
10.52202/079017-0934
10.52202/079017-0934 · External reference
Risks from learned optimization in advanced machine learning systems
2019 · External reference
Process supervision: Aligning language models via intermediates
2024 · External reference
ToolLLM: Facilitating large language models to master 16,000 realworld apis
2024 · External reference
10.1609/aaai.v39i26.34991
10.1609/aaai.v39i26.34991 · External reference
Rlaif: Reinforcement learning from human feedback to mitigate reward hacking
2023 · External reference
10.1016/j.engappai.2024.108553
10.1016/j.engappai.2024.108553 · External reference
Aligning LLM agents by learning latent preferences from user edits (prelude)
2024 · External reference
Robust CLIP: Unsupervised adversarial fine-tuning of vision embeddings
2024 · External reference
10.1145/3746027.3755646
10.1145/3746027.3755646 · External reference
Survey of adversarial robustness in multimodal large language models
2025 · External reference
10.52202/085713-0593
10.52202/085713-0593 · External reference
MINI-LLM: Memory-efficient structured pruning for large language models
2024 · External reference
Dissecting adversarial robustness of multimodal LM agents
2024 · External reference
10.1145/3627673.3679222
10.1145/3627673.3679222 · External reference
10.52202/075280-1512
10.52202/075280-1512 · External reference
10.52202/085713-3801
10.52202/085713-3801 · External reference
ToolLLM: Facilitating large language models to master 16000 real-world apis
2024 · External reference
Constrained reinforcement learning has zero duality gap
2019 · External reference
LLM powered autonomous agents
2023 · External reference
Shieldagent: Shielding agents via verifiable safety policy reasoning
2025 · External reference
The fusion of large language models and formal methods for trustworthy AI agents: A roadmap
2024 · External reference
Stair: Improving safety alignment with introspective reasoning
2025 · External reference
Realsafe-r1: Safety-aligned deepseek-r1 without compromising reasoning capability
2025 · External reference
10.18653/v1/2024.emnlp-main.861
10.18653/v1/2024.emnlp-main.861 · External reference
AegisLLM: Scaling agentic systems for selfreflective defense in LLM security
2025 · External reference
Reflective multiagent collaboration based on large language models
2024 · External reference
MAEBE: Multi-agent emergent behavior evaluation framework
2025 · External reference
