1generatedAt: '2026-09-03T16:20:17.969Z'
2highlighted:
3 - slug: amodei-2016-concrete-problems-ai-safety
4 name: Concrete Problems in AI Safety
5 year: 2016
6 arxiv: '1606.06565'
7 venue: null
8 citedByCount: null
9 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1606.06565/2026-08-24/22_24_02_294/artifacts/postprocess/convert/tex-to-markdown.md
10 highlight: Predecessor to G4
11 - slug: park-2023-generative-agents-interactive-simulacra
12 name: 'Generative Agents: Interactive Simulacra of Human Behavior'
13 year: 2023
14 arxiv: '2304.03442'
15 venue: uist
16 citedByCount: 179
17 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2304.03442/2026-08-24/18_18_50_769/artifacts/postprocess/convert/tex-to-markdown.md
18 highlight: 'Smallville: model paper for Agentic Life/Society'
19 - slug: schlatter-2025-incomplete-tasks-induce-shutdown
20 name: Incomplete Tasks Induce Shutdown Resistance in Some Frontier LLMs
21 year: 2026
22 arxiv: '2509.14260'
23 venue: venue/tmlr
24 citedByCount: 0
25 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2509.14260/2026-08-22/13_05_01_501/artifacts/postprocess/convert/tex-to-markdown.md
26 highlight: '"The agent was just confused" - self-preserving shutdown dissolved when priorities clarified.'
27 - slug: shah-2025-approach-technical-agi-safety
28 name: An Approach to Technical AGI Safety and Security
29 year: 2025
30 arxiv: '2504.01849'
31 venue: null
32 citedByCount: null
33 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2504.01849/2026-08-21/18_20_41_030/artifacts/postprocess/convert/tex-to-markdown.md
34 highlight: '"Google 4" Risk Areas: Misuse, Misalignment, Mistakes, Structural Risks'
35 - slug: singh-2026-model-forensics-investigating-whether
36 name: 'Model Forensics: Investigating Whether Concerning Behavior Reflects Misalignment'
37 year: 2026
38 arxiv: '2606.26071'
39 venue: null
40 citedByCount: 0
41 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2606.26071/2026-08-19/15_52_12_656/artifacts/postprocess/convert/tex-to-markdown.md
42 highlight: Companion blog post has 10 examples of "Benign Explanations"
43enhanced:
44 - slug: adversarial-paraphrasing
45 name: 'Adversarial Paraphrasing: A Universal Attack for Humanizing AI-Generated Text'
46 year: 2025
47 arxiv: '2506.07001'
48 venue: null
49 citedByCount: 0
50 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2506.07001/2026-08-18/14_38_24_546/artifacts/postprocess/convert/tex-to-markdown.md
51 - slug: akrout-2019-v3
52 name: Hacking Google reCAPTCHA v3 using Reinforcement Learning
53 year: 2019
54 arxiv: '1903.01003'
55 venue: null
56 citedByCount: 15
57 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1903.01003/2026-08-13/22_57_59_324/artifacts/postprocess/convert/tex-to-markdown.md
58 - slug: arjovsky-2017-principled
59 name: Towards Principled Methods for Training Generative Adversarial Networks
60 year: 2017
61 arxiv: '1701.04862'
62 venue: iclr
63 citedByCount: 357
64 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1701.04862/2026-08-13/22_54_32_172/artifacts/postprocess/convert/tex-to-markdown.md
65 - slug: arjovsky-2017-wgan
66 name: Wasserstein GAN
67 year: 2017
68 arxiv: '1701.07875'
69 venue: icml
70 citedByCount: 621
71 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1701.07875/2026-08-13/22_54_45_312/artifacts/postprocess/convert/tex-to-markdown.md
72 - slug: athalye-2018-obfuscated
73 name: 'Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples'
74 year: 2018
75 arxiv: '1802.00420'
76 venue: icml
77 citedByCount: 1158
78 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1802.00420/2026-08-13/22_56_43_562/artifacts/postprocess/convert/tex-to-markdown.md
79 - slug: baker-2020-hide-and-seek
80 name: Emergent Tool Use From Multi-Agent Autocurricula
81 year: 2019
82 arxiv: '1909.07528'
83 venue: iclr
84 citedByCount: 333
85 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1909.07528/2026-08-13/22_59_12_617/artifacts/postprocess/convert/tex-to-markdown.md
86 - slug: balduzzi-2018-mechanics
87 name: The Mechanics of n-Player Differentiable Games
88 year: 2018
89 arxiv: '1802.05642'
90 venue: icml
91 citedByCount: 82
92 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1802.05642/2026-08-13/22_56_57_373/artifacts/postprocess/convert/tex-to-markdown.md
93 - slug: balduzzi-2019-open-ended
94 name: Open-ended Learning in Symmetric Zero-sum Games
95 year: 2019
96 arxiv: '1901.08106'
97 venue: icml
98 citedByCount: 46
99 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1901.08106/2026-08-13/22_57_23_909/artifacts/postprocess/convert/tex-to-markdown.md
100 - slug: bao-2024-fast-detectgpt
101 name: 'Fast-DetectGPT: Efficient Zero-Shot Detection of Machine-Generated Text via Conditional Probability Curvature'
102 year: 2023
103 arxiv: '2310.05130'
104 venue: iclr
105 citedByCount: 19
106 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2310.05130/2026-08-13/23_06_32_761/artifacts/postprocess/convert/tex-to-markdown.md
107 - slug: base-models-look-human
108 name: Base Models Look Human To AI Detectors
109 year: 2026
110 arxiv: '2605.19516'
111 venue: null
112 citedByCount: 0
113 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2605.19516/2026-08-18/14_47_26_718/artifacts/postprocess/convert/tex-to-markdown.md
114 - slug: becaptcha-mouse-2022
115 name: 'BeCAPTCHA-Mouse: Synthetic Mouse Trajectories and Improved Bot Detection'
116 year: 2022
117 arxiv: '2005.00890'
118 venue: null
119 citedByCount: 42
120 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2005.00890/2026-08-13/23_00_31_181/artifacts/postprocess/convert/tex-to-markdown.md
121 - slug: bengio-2025-international-ai-safety-report
122 name: International AI Safety Report
123 year: 2025
124 arxiv: '2501.17805'
125 venue: null
126 citedByCount: null
127 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2501.17805/2026-08-24/22_26_50_098/artifacts/postprocess/convert/pdf-to-markdown.md
128 - slug: bengio-2026-international-ai-safety-report
129 name: International AI Safety Report 2026
130 year: 2026
131 arxiv: '2602.21012'
132 venue: null
133 citedByCount: null
134 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2602.21012/2026-08-24/22_32_40_114/artifacts/postprocess/convert/pdf-to-markdown.md
135 - slug: bias-nonnative
136 name: GPT detectors are biased against non-native English writers
137 year: 2023
138 arxiv: '2304.02819'
139 venue: null
140 citedByCount: 40
141 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2304.02819/2026-08-13/23_03_46_060/artifacts/postprocess/convert/tex-to-markdown.md
142 - slug: binoculars
143 name: 'Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Text'
144 year: 2024
145 arxiv: '2401.12070'
146 venue: null
147 citedByCount: 16
148 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2401.12070/2026-08-13/23_07_34_064/artifacts/postprocess/convert/tex-to-markdown.md
149 - slug: broken-gates-2026
150 name: 'Broken Gates: Re-evaluating Web Bot Defenses in the Age of LLM Agents'
151 year: 2026
152 arxiv: '2607.18659'
153 venue: null
154 citedByCount: 0
155 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/2607.18659/2026-08-13/23_14_37_790/artifacts/postprocess/convert/tex-to-markdown.md
156 - slug: caccia-2020-falling-short
157 name: Language GANs Falling Short
158 year: 2018
159 arxiv: '1811.02549'
160 venue: iclr
161 citedByCount: 81
162 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1811.02549/2026-08-13/22_57_11_250/artifacts/postprocess/convert/tex-to-markdown.md
163 - slug: carlini-2017-cw
164 name: Towards Evaluating the Robustness of Neural Networks
165 year: 2016
166 arxiv: '1608.04644'
167 venue: null
168 citedByCount: 661
169 markdown: /Users/eshao/.config/mnt/runs/query-arxiv/1608.04644/2026-08-13/22_53_58_274/artifacts/postprocess/convert/tex-to-markdown.md
170 - slug: chen-2024-spin
171 name: Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models
172 year: