Pith. sign in

REVIEW 3 major objections 4 minor 51 references

AI Safety vs. AI Security: Demystifying the Distinction and Boundaries

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read AI safety vs. AI security: intent draws the line

desk verdict Useful, clearly-written conceptual synthesis, but the claimed precision overreaches because the paper never pins down whose intent distinguishes safety from security. read the letter →

arxiv 2506.18932 v1 pith:AHUFZETC submitted 2025-06-21 cs.CY cs.AIcs.CR

Zhiqiang Lin , Huan Sun , Ness Shroff This is my paper · ORCID
classification cs.CYcs.AIcs.CR
keywords AIsafetysecurityintentionalvsunintentionalharmadversarialattacksrisktaxonomytrustworthymisuse
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that AI Safety and AI Security are not interchangeable terms but distinct research domains separated by the origin of the risk: AI Safety addresses accidental, unintended harm from system flaws, misalignment, or distributional shifts, while AI Security addresses intentional adversarial actions such as data poisoning, evasion attacks, prompt injection, and model theft. The authors propose formal definitions of both concepts, show how they are interdependent, and argue that keeping them distinct is essential for focused research, effective regulation, and trustworthy AI. If the distinction holds, it gives regulators and funders a clean way to route risks to the right mitigation tools.

What carries the argument

The argument is carried by an intent-based dichotomy formalized in two named definitions: Definition 1 for AI Safety and Definition 2 for AI Security. These definitions do the classificatory work throughout the paper, supported by two analogies: the communication model where a checksum (CRC) detects accidental corruption while a message authentication code (MAC) resists deliberate tampering, and the building analogy where structural integrity maps to safety while locks, alarms, and perimeter defenses map to security. The intent axis is used to classify research topics, case studies, and the relationship between the two domains.

What would settle it

A documented AI incident in which the same harmful output arises simultaneously from an unintended design flaw and a deliberate exploit, and where a panel of safety and security experts cannot agree on which category it belongs to, would show the intent axis is not precise enough to carry the claimed boundary.

Watch

Extended reading notes

Core claim

The paper's central claim is that the primary distinction between AI Safety and AI Security lies in the origin and nature of the risk: safety concerns accidental or unintended behaviors, while security concerns intentional adversarial actions. Definition 1 defines AI Safety as the property of avoiding unintended harmful outcomes despite uncertainties in inputs, goals, training data, or deployment conditions. Definition 2 defines AI Security as the property of remaining resilient against intentional attacks on data, algorithms, or operations, preserving confidentiality, integrity, and availability. The paper argues that misuse cases often sit at the intersection of these two categories because safety flaws can be exploited by adversaries, and security breaches can produce safety failures, yet the intent-based boundary is still the correct organizing axis for the field.

Load-bearing premise

The paper assumes that harmful AI events can be reliably classified by whether the cause was accidental or intentional, even though it concedes that many misuse cases sit at the intersection of the two.

Editorial extensions

If this is right

  • Research funding and agendas can be split by threat origin: test-and-align work goes to safety; adversarial robustness, access control, and incident response go to security.
  • Policy tools can be matched to risk type: pre-deployment testing and ethical standards for safety; cybersecurity standards, red teaming, and breach notification for security.
  • Systems need both properties: a safe but insecure system can be hacked, while a secure but unsafe system can still harm through bias or misalignment.
  • Unified AI risk management must trace chains that cross the boundary, such as a security attack like prompt injection that defeats a safety filter and produces harmful output.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable prediction: expert disagreement over classifying ambiguous incidents, such as a model that generates harmful content both because of training-data bias and because of a jailbreak, will persist, suggesting the boundary is a spectrum rather than a clean partition.
  • A practical extension: AI incident reporting could tag every event on two independent axes—cause (accidental vs. intentional) and harm type—so that the field inherits the safety/security separation that aviation and nuclear power have long used.
  • The intent axis could be operationalized empirically: if an incident is fully resolved by retraining or re-specification it leans safety; if it is resolved only by authentication, filtering, or access controls it leans security.
  • The paper's framework implies that a single AI system may require separate assurance cases for safety and security, each with its own evidence and ownership, which current practice rarely distinguishes.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This paper is a conceptual/taxonomic contribution arguing that AI Safety and AI Security should be distinguished by the origin and nature of the risk. It proposes intent-based definitions (Definition 1: AI Safety as avoiding unintended harmful outcomes despite uncertainties; Definition 2: AI Security as resilience against intentional attacks on data, algorithms, or operations), reviews the historical broadening of the term "AI Safety," illustrates the distinction with message-transmission and building analogies, maps distinct research agendas and protective mechanisms, and concludes that the two domains are interdependent and should be managed within a unified risk framework. The authors explicitly acknowledge in Section 4.2 that misuse often lies at the intersection of unintentional system flaws and intentional exploitation.

Significance. If the proposed taxonomy could be made precise, it would help align research agendas, funding, and regulation with distinct classes of AI risk, and the paper is a readable synthesis of the ongoing safety-security debate. Its strengths are a clear organization, a substantial set of references to relevant literature, and an honest acknowledgment of edge cases. As it stands, however, the claimed "precise boundaries" are not supported because the central notion of intent is ambiguous and the two definitions are not mutually exclusive for the misuse cases the paper itself identifies. The contribution is therefore currently a useful heuristic rather than a rigorous delineation.

major comments (3)
  1. [§3, Definitions 1 and 2; §4.1] The central distinction rests on intent, but the manuscript never states whose intent is the relevant one. Definition 1's "unintended" is ambiguous among the AI system's own goals, the developer's or operator's intent, and the user's intent, while Definition 2's "intentional attacks" refers to an adversary. These readings diverge: a power-seeking AI (Section 5.1, reference [14]) deliberately resists shutdown, so under a system-intent reading it is not a safety failure; conversely, a user deliberately exploiting a known bias (the kind of case mentioned in Section 4.2) is unintended by the developer but intentionally caused by the user, so the same event satisfies both definitions. The paper needs an explicit attribution rule for intent, or a classification procedure, before the claimed "precise boundary" can be supported.
  2. [§4.2] The paper concedes that "misuse often lies at the intersection" of unintentional system flaws and intentional exploitation, yet provides no criterion for resolving such cases. Definitions 1 and 2 are not mutually exclusive as stated: a prompt-injection attack that succeeds because of an alignment flaw is simultaneously an unintended harmful outcome of the system and an intentional attack by an adversary. Section 4.1's examples do not cover this case, so the proposed boundary cannot classify the central phenomenon—AI misuse—that the paper says it clarifies.
  3. [§3.1 and §5.1] Section 5.1 lists "power-seeking behaviors" and "misaligned objectives" under AI Safety, but these are goal-directed behaviors rather than accidental system failures. This conflicts with Definition 1's characterization of safety as preventing "unintended harmful outcomes" and indicates a conflation between "not intended by the designers" and "accidental system malfunction." The authors should clarify that safety includes failures of alignment even when the system competently pursues a misaligned goal, or they should restrict the definition accordingly.
minor comments (4)
  1. [§7] The word "illustratred" should be "illustrated."
  2. [§6] In the paragraph on autonomous vehicles, "A Vs" should be "AVs."
  3. [§3.2] The definition attributed to "Bengio et al." is a quotation from the International AI Safety Report [6], which is an institutional multi-author report; the attribution should be clarified.
  4. [§1] The message-transmission analogy treats safety failures as stochastic noise, but many AI safety failures, such as misgeneralization or goal misalignment, are not well modeled by random corruption; the analogy should be explicitly labeled as heuristic only.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the intent-based boundary is a stipulated definition, not a derived prediction.

full rationale

The paper is a conceptual synthesis, not an empirical derivation. Its central distinction (Section 4.1) restates the definitions proposed in Section 3, but this is by design and is explicitly labeled a proposal: the paper says 'we propose the following precise definitions.' It does not fit parameters, make quantitative predictions, invoke a uniqueness theorem, or rely on self-citations; the cited sources are independent external works. The message-transmission and building analogies are illustrative rather than load-bearing. The only potential concern is that the boundary is definitional, and Section 4.2 itself concedes that 'misuse often lies at the intersection' of unintentional system flaws and intentional exploitation, but that is a limitation of the proposed taxonomy's precision, not circularity. A taxonomy paper that stipulates its axis and then organizes the field along that axis is not deriving its conclusion from its inputs in the problematic sense; the conclusion is the announced consequence of the stipulated axis. Therefore no circular step is present.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No quantities are fitted and no new entities are postulated. The central claim rests on three conceptual assumptions: the engineering definitions transfer to AI, intent is a useful practical classifier, and the historical separation of the two communities is the right motivation for separate concepts.

assumptions (3)
  • domain assumption The classical engineering definitions of safety as protection from unintentional harm and security as protection from intentional harm transfer directly to AI systems.
    Invoked at the start of Section 3; establishes the conceptual foundation for Definitions 1 and 2 using Leveson [30] and Anderson [3].
  • ad hoc to paper Intent is a meaningful and practically determinable criterion for classifying AI risk.
    This is the paper's load-bearing premise; Section 4.2 concedes that misuse often lies at the intersection of safety and security, which makes intent harder to determine in practice.
  • domain assumption The historical separation of AI safety and cybersecurity research communities is an appropriate reason to maintain distinct definitions.
    The paper relies on Qi et al. [39] to justify why the boundary should be drawn, as discussed in Section 2.2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AI Safety vs. AI Security: Demystifying the Distinction and Boundaries." pith.science (2026). https://pith.science/paper/AHUFZETC

@misc{pith2026250618932,
  author       = {Pith},
  title        = {Pith review of: AI Safety vs. AI Security: Demystifying the Distinction and Boundaries},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AHUFZETC}},
  note         = {Machine review of arXiv:2506.18932}
}
read the original abstract

Artificial Intelligence (AI) is rapidly being integrated into critical systems across various domains, from healthcare to autonomous vehicles. While its integration brings immense benefits, it also introduces significant risks, including those arising from AI misuse. Within the discourse on managing these risks, the terms "AI Safety" and "AI Security" are often used, sometimes interchangeably, resulting in conceptual confusion. This paper aims to demystify the distinction and delineate the precise research boundaries between AI Safety and AI Security. We provide rigorous definitions, outline their respective research focuses, and explore their interdependency, including how security breaches can precipitate safety failures and vice versa. Using clear analogies from message transmission and building construction, we illustrate these distinctions. Clarifying these boundaries is crucial for guiding precise research directions, fostering effective cross-disciplinary collaboration, enhancing policy effectiveness, and ultimately, promoting the deployment of trustworthy AI systems.

Figures

Figures reproduced from arXiv: 2506.18932 by the authors.

Figure 1
Figure 1. Conceptual Essence of AI Safety: Focuses on preventing unintended harm from internal [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Conceptual Essence of AI Security: Focuses on protecting the AI system from intentional, [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Distinctions and Interplay of AI Safety and AI Security. Misuse often lies at the intersection, [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Building analogy showing AI Safety (structural integrity) versus AI Security (protection [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Conceptual Framework for Unified AI Risk Management, integrating AI Safety and AI [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 32 canonical work pages

  1. [14]

    Is power-seeking ai an existential risk?arXiv preprint arXiv:2206.13353, 2022

    Joseph Carlsmith. Is power-seeking ai an existential risk?arXiv preprint arXiv:2206.13353, 2022

    arXiv 2022

  2. [1]

    The role of machine and deep learning in modern intrusion detection systems: A comprehensive review.Computers and Electrical Engineering, 124:110318, 2025

    Uday Chandra Akuthota and Lava Bhargava. The role of machine and deep learning in modern intrusion detection systems: A comprehensive review.Computers and Electrical Engineering, 124:110318, 2025. ISSN 0045-7906. doi: https://doi.org/10.1016/j.compeleceng.2025.110318

    arXiv 2025

  3. [2]

    Christiano, John Schulman, and Dan Mané

    Dario Amodei, Chris Olah, Jacob Steinhardt, Paul F. Christiano, John Schulman, and Dan Mané. Concrete problems in ai safety.arXiv preprint arXiv:1606.06565, 2016

    arXiv 2016

  4. [3]

    Anderson.Security Engineering: A Guide to Building Dependable Distributed Systems

    Ross J. Anderson.Security Engineering: A Guide to Building Dependable Distributed Systems. Wiley, 3rd edition, 2020

    work page 2020

  5. [4]

    Constitutional ai: Harmlessness from ai feedback.arXiv preprint arXiv:2212.08073, 2022

    Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, Jackson Kernion, Andy Jones, Anna Chen, Anna Goldie, Azalia Mirhoseini, Cameron McKinnon, et al. Constitutional ai: Harmlessness from ai feedback.arXiv preprint arXiv:2212.08073, 2022

    arXiv 2022

  6. [5]

    fairmlbook.org, 2019.http://www.fairmlbook.org

    Solon Barocas, Moritz Hardt, and Arvind Narayanan.Fairness and Machine Learning: Limitations and Opportunities. fairmlbook.org, 2019.http://www.fairmlbook.org

    work page 2019

  7. [6]

    International AI safety report: The international scientific report on the safety of advanced AI

    Yoshua Bengio et al. International AI safety report: The international scientific report on the safety of advanced AI. Technical report, Produced with support from the UK Government, for the AI Safety Summit initiatives, January 2025

    work page 2025

  8. [7]

    Poisoning attacks against support vector machines

    Battista Biggio, Blaine Nelson, and Pavel Laskov. Poisoning attacks against support vector machines. InProceedings of the 29th International Conference on Machine Learning (ICML), 2012

    work page 2012

Show all 51 references
  1. [8]

    On the opportunities and risks of foundation models.arXiv preprint arXiv:2108.07258, 2021

    Rishi Bommasani, Drew A Hudson, Ehsan Adeli, Russ Altman, Simran Arora, Sydney von Arx, Michael S Bernstein, Jeannette Bohg, Antoine Bosselut, Emma Brunskill, et al. On the opportunities and risks of foundation models.arXiv preprint arXiv:2108.07258, 2021

    2021 arXiv

  2. [9]

    The social dilemma of autonomous vehicles.Science, 352(6293):1573–1576, 2016

    Jean-François Bonnefon, Azim Shariff, and Iyad Rahwan. The social dilemma of autonomous vehicles.Science, 352(6293):1573–1576, 2016. doi: 10.1126/science.aaf2654

    2016 doi

  3. [10]

    Oxford University Press, 2016

    Nick Bostrom.Superintelligence: Paths, Dangers, Strategies. Oxford University Press, 2016

    2016

  4. [11]

    Miles Brundage, Shahar Avin, Jasmine Wang, Haydn Belfield, Gretchen Krueger, Gillian Hadfield, Heidy Khlaaf, Jingying Yang, Helen Toner, Ruth Fong, Tegan Maharaj, Pang Wei Koh, Sara Hooker, Jade Leung, Andrew Trask, Emma Bluemke, Cullen O’Keefe, Mark Koren, Théo Ryffel, J. B. ...

    2004 arXiv

  5. [12]

    Towards evaluating the robustness of neural networks

    Nicholas Carlini and David Wagner. Towards evaluating the robustness of neural networks. In Proceedings of the IEEE Symposium on Security and Privacy (S&P), 2017

    2017

  6. [13]

    Extracting training data from large language models

    Nicholas Carlini, Florian Tramer, Eric Wallace, Matthew Jagielski, Ariel Herbert-V oss, Katherine Lee, Adam Roberts, Tom Brown, Dawn Song, Ulfar Erlingsson, et al. Extracting training data from large language models. In30th USENIX security symposium, pages 2633–2650, 2021

    2021

  7. [15]

    Privacy by design: The 7 foundational principles.Information and privacy commissioner of Ontario, Canada, 2009

    Ann Cavoukian. Privacy by design: The 7 foundational principles.Information and privacy commissioner of Ontario, Canada, 2009

    2009

  8. [16]

    Finlayson, John D

    Samuel G. Finlayson, John D. Bowers, Joichi Ito, Jonathan L. Zittrain, Andrew L. Beam, and Isaac S. Kohane. Adversarial attacks on medical machine learning.Science, 363(6433): 1287–1289, 2019. doi: 10.1126/science.aaw4399

    2019 doi

  9. [17]

    Ai4people—an ethical framework for a good ai society: opportunities, risks, principles, and recommendations

    Luciano Floridi, Josh Cowls, Monica Beltrametti, Raja Chatila, Patrice Chazerand, Virginia Dignum, Christoph Luetge, Robert Madelin, Ugo Pagallo, Francesca Rossi, et al. Ai4people—an ethical framework for a good ai society: opportunities, risks, principles, and recommendations...

    2018

  10. [18]

    Model inversion attacks that exploit confidence information and basic countermeasures

    Matt Fredrikson, Somesh Jha, and Thomas Ristenpart. Model inversion attacks that exploit confidence information and basic countermeasures. InProceedings of the 22nd ACM SIGSAC conference on computer and communications security, pages 1322–1333, 2015

    2015

  11. [19]

    Artificial intelligence, values, and alignment.Minds and Machines, 30:411–437, 2020

    Iason Gabriel. Artificial intelligence, values, and alignment.Minds and Machines, 30:411–437, 2020

    2020

  12. [20]

    Red teaming language models to reduce harms: Methods, scaling behaviors, and lessons learned.arXiv preprint arXiv:2209.07858, 2022

    Deep Ganguli, Liane Lovitt, Jackson Kernion, Amanda Askell, Yuntao Bai, Saurav Kadavath, Ben Mann, Ethan Perez, Nicholas Schiefer, Kamal Ndousse, et al. Red teaming language models to reduce harms: Methods, scaling behaviors, and lessons learned.arXiv preprint arXiv:2209.07858, 2022

    2022 arXiv

  13. [21]

    Goodfellow, Jonathon Shlens, and Christian Szegedy

    Ian J. Goodfellow, Jonathon Shlens, and Christian Szegedy. Explaining and harnessing adversarial examples.arXiv preprint arXiv:1412.6572, 2014

    2014 arXiv

  14. [22]

    Not what you’ve signed up for: Compromising real-world llm-integrated applications with indirect prompt injection

    Kai Greshake, Sahar Abdelnabi, Shailesh Mishra, Christoph Endres, Thorsten Holz, and Mario Fritz. Not what you’ve signed up for: Compromising real-world llm-integrated applications with indirect prompt injection. InProceedings of the 16th ACM Workshop on Artificial Intelligenc...

    2023

  15. [23]

    XAI—explainable artificial intelligence.Science Robotics, 4(37), 2019

    David Gunning, Mark Stefik, Jaesik Choi, Timothy Miller, Simone Stumpf, and Guang-Zhong Yang. XAI—explainable artificial intelligence.Science Robotics, 4(37), 2019. doi: 10.1126/ scirobotics.aay7120

    2019

  16. [24]

    Natural adversarial examples.arXiv preprint arXiv:1907.07174, 2019

    Dan Hendrycks, Kevin Zhao, Steven Basart, Jacob Steinhardt, and Dawn Song. Natural adversarial examples.arXiv preprint arXiv:1907.07174, 2019

    1907 arXiv

  17. [25]

    Unsolved problems in ml safety, 2022

    Dan Hendrycks, Nicholas Carlini, John Schulman, and Jacob Steinhardt. Unsolved problems in ml safety, 2022. URLhttps://arxiv.org/abs/2109.13916

    2022 arXiv

  18. [26]

    An overview of catastrophic ai risks

    Dan Hendrycks, Mantas Mazeika, and Thomas Woodside. An overview of catastrophic ai risks. arXiv preprint arXiv:2306.12001, 2023

    2023 arXiv

  19. [27]

    Chapman and hall/CRC, 2007

    Jonathan Katz and Yehuda Lindell.Introduction to modern cryptography: principles and protocols. Chapman and hall/CRC, 2007

    2007

  20. [28]

    Philip Koopman.How safe is safe enough?: Measuring and predicting Autonomous Vehicle Safety. 2022. 12

    2022

  21. [29]

    Experimental security analysis of a modern automobile

    Karl Koscher, Alexei Czeskis, Franziska Roesner, Shwetak Patel, Tadayoshi Kohno, Stephen Checkoway, Damon McCoy, Brian Kantor, Danny Anderson, Hovav Shacham, and Stefan Savage. Experimental security analysis of a modern automobile. InProceedings of the IEEE Symposium on Securi...

    2010

  22. [30]

    Leveson.Engineering a Safer World: Systems Thinking Applied to Safety

    Nancy G. Leveson.Engineering a Safer World: Systems Thinking Applied to Safety. The MIT Press, 2012. doi: 10.7551/mitpress/8179.001.0001

    2012 doi

  23. [31]

    Lundberg and Su-In Lee

    Scott M. Lundberg and Su-In Lee. A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 2017

    2017

  24. [32]

    Towards deep learning models resistant to adversarial attacks.International Conference on Learning Representations (ICLR), 2018

    Aleksander M ˛ adry, Aleksandar Makelov, Ludwig Schmidt, Dimitris Tsipras, and Adrian Vladu. Towards deep learning models resistant to adversarial attacks.International Conference on Learning Representations (ICLR), 2018

    2018

  25. [33]

    Vintage Books, 2019

    Gary Marcus and Ernest Davis.Rebooting AI: Building Artificial Intelligence We Can Trust. Vintage Books, 2019

    2019

  26. [34]

    A survey on bias and fairness in machine learning.ACM Computing Surveys, 54(6):1–35, 2021

    Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. A survey on bias and fairness in machine learning.ACM Computing Surveys, 54(6):1–35, 2021

    2021

  27. [35]

    Security and privacy controls for information systems and organizations

    National Institute of Standards and Technology. Security and privacy controls for information systems and organizations. Technical Report Revision 5, U.S. Department of Commerce, September 2020

    2020

  28. [36]

    Artificial intelligence risk management framework (ai rmf 1.0)

    National Institute of Standards and Technology. Artificial intelligence risk management framework (ai rmf 1.0). Technical report, U.S. Department of Commerce, 2023

    2023

  29. [37]

    Safely interruptible agents.Proceedings of the Thirty-Second Conference on Uncertainty in Artificial Intelligence (UAI), 2016

    Laurent Orseau and Stuart Armstrong. Safely interruptible agents.Proceedings of the Thirty-Second Conference on Uncertainty in Artificial Intelligence (UAI), 2016

    2016

  30. [38]

    Training language models to follow instructions with human feedback.Advances in neural information processing systems, 35:27730–27744, 2022

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback.Advances in neural information processing systems, 35:27730...

    2022

  31. [39]

    Su, Mengdi Wang, Chaowei Xiao, Bo Li, Dawn Song, Peter Henderson, and Prateek Mittal

    Xiangyu Qi, Yangsibo Huang, Yi Zeng, Edoardo Debenedetti, Jonas Geiping, Luxi He, Kaixuan Huang, Udari Madhushani, Vikash Sehwag, Weijia Shi, Boyi Wei, Tinghao Xie, Danqi Chen, Pin-Yu Chen, Jeffrey Ding, Ruoxi Jia, Jiaqi Ma, Arvind Narayanan, Weijie J. Su, Mengdi Wang, Chaowei...

    2024

  32. [40]

    why should i trust you?

    Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin. “why should i trust you?”: Explaining the predictions of any classifier.Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1135–1144, 2016. doi: 10.1145/2939672.2939778

    2016

  33. [41]

    Artificial intelligence and the problem of control.Perspectives on digital humanism, 19:1–322, 2022

    Stuart Russell. Artificial intelligence and the problem of control.Perspectives on digital humanism, 19:1–322, 2022

    2022

  34. [42]

    Ronny Huang, Mahyar Najibi, Octavian Suciu, Christoph Studer, Tudor Dumitras, and Tom Goldstein

    Ali Shafahi, W. Ronny Huang, Mahyar Najibi, Octavian Suciu, Christoph Studer, Tudor Dumitras, and Tom Goldstein. Poison frogs! targeted clean-label poisoning attacks on neural networks.Advances in Neural Information Processing Systems, 31, 2018

    2018

  35. [43]

    Membership inference attacks against machine learning models

    Reza Shokri, Marco Stronati, Congzheng Song, and Vitaly Shmatikov. Membership inference attacks against machine learning models. InProceedings of the IEEE Symposium on Security and Privacy (S&P), pages 3–18, 2017

    2017

  36. [44]

    Secure software development framework (ssdf) version 1.1.NIST Special Publication, 800(218):800–218, 2022

    Murugiah Souppaya, Karen Scarfone, and Donna Dodson. Secure software development framework (ssdf) version 1.1.NIST Special Publication, 800(218):800–218, 2022

    2022

  37. [45]

    Tanenbaum and David J

    Andrew S. Tanenbaum and David J. Wetherall.Computer Networks. Prentice Hall, 5th edition, 2010. 13

    2010

  38. [46]

    Ai security has serious terminology issues

    Joseph Thacker. Ai security has serious terminology issues. URL https://josephthacker. com/ai/2023/10/16/ai-security-terminology-issues.html

    2023

  39. [47]

    Hachette UK, 2019

    Eric Topol.Deep medicine: how artificial intelligence can make healthcare human again. Hachette UK, 2019

    2019

  40. [48]

    Reiter, and Thomas Ristenpart

    Florian Tramèr, Fan Zhang, Ari Juels, Michael K. Reiter, and Thomas Ristenpart. Stealing machine learning models via prediction APIs. InProceedings of the 25th USENIX Security Symposium, pages 601–618, 2016

    2016

  41. [49]

    Embedding watermarks into deep neural networks

    Yusuke Uchida, Yuki Nagai, Shigeyuki Sakazawa, and Shin’ichi Satoh. Embedding watermarks into deep neural networks. InProceedings of the 2017 ACM on international conference on multimedia retrieval, pages 269–277, 2017

    2017

  42. [50]

    Ethical and social risks of harm from language models.arXiv preprint arXiv:2112.04359, 2021

    Laura Weidinger, John Mellor, Maribeth Rauh, Conor Griffin, Jonathan Uesato, Po-Sen Huang, Myra Cheng, Mia Glaese, Borja Balle, Atoosa Kasirzadeh, et al. Ethical and social risks of harm from language models.arXiv preprint arXiv:2112.04359, 2021

    2021 arXiv

  43. [51]

    Improving alignment and robustness with circuit breakers, 2024

    Andy Zou, Long Phan, Justin Wang, Derek Duenas, Maxwell Lin, Maksym Andriushchenko, Rowan Wang, Zico Kolter, Matt Fredrikson, and Dan Hendrycks. Improving alignment and robustness with circuit breakers, 2024. URLhttps://arxiv.org/abs/2406.04313. 14

    2024 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.