Pith. sign in

REVIEW 4 major objections 4 minor 40 references

The Age of Sensorial Zero Trust: Why We Can No Longer Trust Our Senses

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper argues that because generative AI can now fabricate faces, voices, and video that fool humans, organizations must stop treating sensory information as trustworthy by default, extending the Zero Trust security principle "never…

desk verdict A readable practitioner's brief that renames standard zero-trust controls 'Sensorial Zero Trust'; the human-detection pillar is contradicted by its own cited evidence, but the rest of the package is sound. read the letter →

arxiv 2507.00907 v1 pith:OFP52LXI submitted 2025-07-01 cs.CR cs.AI

Fabio Correa Xavier This is my paper
classification cs.CRcs.AI
keywords zerotrustdeepfakedetectionvoicecloningout-of-bandverificationmultifactorauthenticationsensorysocialengineeringgenerativeAIfraud
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 because generative AI can now fabricate faces, voices, and video that fool humans, organizations must stop treating sensory information as trustworthy by default. It extends the Zero Trust security principle — "never trust, always verify" — to human perception, proposing that every high-stakes request be confirmed through an independent channel before action. The paper surveys documented deepfake frauds, reports that most people cannot reliably detect cloned voices, and lays out four technical pillars plus training to operationalize this doubt. If the argument is right, standard practices like approving wire transfers on a recognized voice or face become procedures that require explicit, multi-channel verification.

What carries the argument

The load-bearing mechanism is the extension of the Zero Trust cycle — verify explicitly, use least privilege, assume breach — to the human sensory channel, implemented through out-of-band verification: confirming any sensitive request through a second, independent communication channel that an attacker would have to compromise simultaneously. This is supported by extended multi-factor authentication for verbal and video requests, continuous authentication through behavioral biometrics, automated deepfake and liveness detectors, and awareness training that rewards employees for doubting authority. The framework's force depends on the claim that deepfake and voice-clone artifacts, while often invisible to humans, can be caught by either a second channel or machine analysis.

What would settle it

Compare fraud rates at similar organizations that do and do not adopt mandatory out-of-band verification for high-value transfers over a fixed period; if impersonation fraud is not measurably lower in the adopting group, the central practical claim breaks. A narrower test is to present trained employees with realistic deepfake calls and measure whether the proposed verification steps catch more fakes than untrained employees while keeping false alarms at an acceptable level.

Watch

Extended reading notes

Core claim

The central claim is that verifiable sensory input has become a security boundary, and the same Zero Trust logic applied to networks must be applied to human perception: no image, voice, or video should be accepted as genuine without evidence. The paper introduces Sensorial Zero Trust as a framework that grafts four controls — out-of-band verification, extended multi-factor authentication, continuous behavioral biometrics, and automated deepfake detection — onto existing enterprise security, supported by training that normalizes constructive doubt. It grounds this in three kinds of evidence: documented impersonation frauds (a Hong Kong deepfake video conference that moved US$25.6 million, a US$35 million voice-clone bank transfer), aggregate incident statistics (a tenfold rise in detected deepfakes from 2022 to 2023), and perception studies showing listeners misidentify cloned voices as the real person roughly 80% of the time. The paper's recommendation is that "verify your eyes and ears when it matters" becomes standard operating procedure, not paranoia.

Load-bearing premise

The framework assumes that the controls it recommends — out-of-band verification, extended MFA, behavioral biometrics, and automated detection — actually reduce fraud in live organizational settings, yet the paper provides no experimental or field data showing that they do.

Editorial extensions

If this is right

  • Wire transfer and payment authorization policies should require confirmation through a known, pre-registered channel before execution, treating any single-channel request as untrusted.
  • Video conferences involving financial or confidential decisions should be recorded and scanned by automated deepfake and liveness detectors, with any positives triggering human verification.
  • Executives and employees should be trained and authorized to challenge urgent requests, including from leadership, and organizations should run simulated deepfake attacks to measure readiness.
  • Identity and access management should adopt cryptographic provenance for media, so that signed content is trusted and unsigned content is treated as suspect.
  • Incident response plans should include deepfake-specific playbooks for reputational attacks, fake executive communications, and fraudulent payment requests.

Reading between the lines

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

  • An implication not developed in the paper is that the same verification discipline should apply to machine consumers of media, since a deepfake that fools a human can also poison downstream automated decisions.
  • A testable extension would embed out-of-band verification directly into virtual meeting platforms, automatically prompting a second-factor check whenever payment or confidential data is requested; the paper does not specify this implementation.
  • If the argument is adopted broadly, the default expectation of truthfulness in everyday communication would shift toward a verify-before-act norm, which could create measurable organizational friction even in the absence of active attacks.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper argues that AI-generated deepfakes and voice clones make human sensory perception an untrusted channel in organizational settings, and it proposes 'Sensorial Zero Trust' as an extension of conventional Zero Trust principles to visual and auditory information. It reviews the threat landscape, presents several high-profile fraud cases, and outlines four implementation pillars (out-of-band verification, extended multi-factor authentication, continuous authentication and behavioral biometrics, and automated deepfake detection) plus a human-training component. The paper is a position/advocacy piece rather than a report of new empirical or analytical research.

Significance. The topic is timely and practically relevant: deepfake-enabled fraud is a real and growing concern, and the paper usefully compiles case studies and references that organizations can consult. The proposed framework is plausible and aligns with existing best practices such as out-of-band verification and multi-factor authentication. However, the paper's scientific contribution is limited: it presents no new evidence, provides no formal model or evaluation of the proposed controls, and its central practical claims rest on industry statistics and unvalidated assumptions about the effectiveness of the recommended defenses. The paper could serve as a high-level briefing for practitioners, but it does not, in its current form, meet the evidentiary standard of a research article.

major comments (4)
  1. [Section 3.2 and Section 5.5] The paper's own evidence undermines the human-training pillar. Section 3.2 cites Barrington et al. [32] that trained listeners correctly identify AI-cloned voices only about 60% of the time and misattribute identity approximately 80% of the time, noting that even trained individuals are barely better than chance. Yet Section 5.5 prescribes exactly this kind of training—spotting blinking artifacts, audio spectral tells, and asking spontaneous questions—based on generic awareness-training references [8,10] that contain no deepfake-specific efficacy data. No experiment is reported showing such training moves accuracy beyond the ~60% baseline or that it does not simply increase false confidence. Because the human-detection component is load-bearing for the claimed multi-layered defense, the authors should either remove this pillar or substantially reframe it as training in verification procedures (e.g., mandatory OOB checks, MFA challenges) rather than sensory discrimination, while explicitly acknowledging the Barrington result.
  2. [Section 4.2] The threat statistics are presented as established facts but come from industry reports and press summaries with unclear methodologies. For example, the '1000% global increase in detected deepfake incidents from 2022 to 2023' is attributed to Sumsub [34], and the Gartner prediction that 50% of phishing attacks will leverage deepfakes is cited via a trade news item [37]. The paper should either cite the primary peer-reviewed sources if they exist or clearly label these figures as vendor estimates with appropriate uncertainty. This matters because the entire risk narrative and the urgency of the proposed framework rest on these numbers.
  3. [Section 5.4] The effectiveness of the automated detection pillar is asserted without evidence. The text concedes that 'no detector is foolproof' and 'adversaries adapt,' but it gives no quantitative performance data, false-positive or false-negative rates, or results from realistic use in live videoconferencing or call workflows. For a framework whose practical value depends on these controls reducing fraud, the absence of validation means the central claim remains an unsupported assertion rather than a demonstrated result.
  4. [Section 5 overall] The proposed controls (OOB verification, extended MFA, continuous behavioral biometrics) are described only qualitatively, and no field data, case studies with the controls in place, or simulated attack experiments are reported to show that they reduce fraud in real organizational workflows. The abstract and Section 1 promise a 'scientific analysis' of the mitigation approach, but the paper does not provide any empirical or simulated test of its recommendations. To support the central practical claim, at least one of the proposed layers should be tested or supported by evidence that specifically addresses deepfake and voice-clone scenarios, rather than by general zero-trust and authentication literature.
minor comments (4)
  1. [Section 8] Section 8 ('Final Considerations') reproduces Section 7 almost verbatim and contains the broken citation '[2It' where a reference marker is malformed. This duplicate should be removed and replaced with a genuine concluding section that synthesizes the argument and states limitations.
  2. [References] The reference list includes numerous entries never cited in the text (e.g., [11]-[17], [25]-[29]) and several that appear tangential to the topic, such as federated-learning poisoning defenses. The paper should cite only works actually used, and all in-text citations should be formatted consistently.
  3. [Abstract and Section 1] The abstract promises 'Vision-Language Models (VLMs) as forensic collaborators' and 'cryptographic provenance' as key concepts, but these are only briefly mentioned in Section 7 and are not developed as parts of the proposed framework. The abstract should align with the paper's actual coverage.
  4. [General] There are multiple typos and spacing irregularities, including 'Key concept s' in the abstract, '[8 , 10 ]' in the text, and inconsistent spacing around punctuation. A careful proofreading pass is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is a prescriptive framework proposal based on external empirical citations and established Zero Trust principles, with no fitted parameters, equations, or self-citation chain.

full rationale

This is a position/framework paper, not a derivation. It argues from established Zero Trust principles (Kindervag, NIST SP 800-207) and external empirical studies (e.g., Barrington et al. on voice-clone detection, Sumsub deepfake incident data, McAfee/Kaspersky surveys) to the normative conclusion that organizations should verify sensory inputs before acting on them. The conclusion is not defined into the inputs: 'Sensorial Zero Trust' is explicitly introduced as an extension of Zero Trust to human perception, and the paper's own caveats (Section 5.3: continuous authentication is 'still an evolving field'; Section 5.4: 'no detector is foolproof—adversaries adapt') acknowledge the limits of the controls without claiming a mathematical derivation. There is no fitted parameter, no prediction obtained from a fitted subset, and no load-bearing self-citation: the reference list contains no self-citations by the author, and the central citations are to independent peer-reviewed work (Scientific Reports) and industry reports. The skeptic's concern that the human-training pillar (§5.5) is unsupported by efficacy data is a correctness/evidence gap, not circularity, because the paper does not assume the effectiveness of training in order to conclude that verification is needed. Accordingly, no circular step can be exhibited under the hard rules, and the appropriate score is 0.

Assumptions & free parameters 0 free parameters · 3 assumptions · 1 invented entities

The paper is a positional synthesis with no mathematical or empirical contribution. Its free-parameter count is zero because it does not fit anything. The main axioms are the reliability of the cited threat statistics and the transferability of Zero Trust controls to human communication.

assumptions (3)
  • domain assumption Human sensory perception is a critical trust boundary that can be reliably spoofed by generative AI
    The entire framework depends on the premise that deepfakes and voice clones are convincing enough to fool humans; Sections 3 and 4 present cited evidence but no new demonstration.
  • domain assumption Traditional Zero Trust principles extend from network entities to human sensory inputs
    The paper asserts this extension in Sections 1 and 2, but no formal argument or empirical study shows that controls like MFA and OOB verification are effective when applied to communication verification.
  • domain assumption The cited industry statistics (e.g., Sumsub, Gartner, Kaspersky) accurately measure deepfake incidence and impact
    The quantitative risk case in Section 4.2 is built entirely on these secondary sources, which the paper does not critically assess.
invented entities (1)
  • Sensorial Zero Trust
    purpose: A named framework for applying Zero Trust verification principles to human sensory information
    The term is introduced by the author; no empirical evaluation of the framework's effectiveness is provided in the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Age of Sensorial Zero Trust: Why We Can No Longer Trust Our Senses." pith.science (2026). https://pith.science/paper/OFP52LXI

@misc{pith2026250700907,
  author       = {Pith},
  title        = {Pith review of: The Age of Sensorial Zero Trust: Why We Can No Longer Trust Our Senses},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OFP52LXI}},
  note         = {Machine review of arXiv:2507.00907}
}
read the original abstract

In a world where deepfakes and cloned voices are emerging as sophisticated attack vectors, organizations require a new security mindset: Sensorial Zero Trust [9]. This article presents a scientific analysis of the need to systematically doubt information perceived through the senses, establishing rigorous verification protocols to mitigate the risks of fraud based on generative artificial intelligence. Key concepts, such as Out-of-Band verification, Vision-Language Models (VLMs) as forensic collaborators, cryptographic provenance, and human training, are integrated into a framework that extends Zero Trust principles to human sensory information. The approach is grounded in empirical findings and academic research, emphasizing that in an era of AI-generated realities, even our eyes and ears can no longer be implicitly trusted without verification. Leaders are called to foster a culture of methodological skepticism to protect organizational integrity in this new threat landscape.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

40 extracted references · 27 canonical work pages

  1. [33]

    Deepfake: banco perdeu US$ 35 milhões por falsificação da voz de um executivo [Internet]

    IstoÉ Dinheiro (Redação). Deepfake: banco perdeu US$ 35 milhões por falsificação da voz de um executivo [Internet]. IstoÉ Dinheiro; 2021 [accessed 2025 Jan 9]. Available from: https://www.istoedinheiro.com.br/deepfake-banco-perdeu-us-35-milhoes-por-falsificacao-da-voz- de-um-executivo/

    work page 2021

  2. [32]

    People are poorly equipped to detect AI-powered voice clones

    Barrington S, Cooper EA, Farid H. People are poorly equipped to detect AI-powered voice clones. Sci Rep. 2025;15(1):11004. DOI: 10.1038/s41598-025-68109-7. (Preprint arXiv:2410.03791 v2, last revised 2025 Jan 25.)

    arXiv 2025

  3. [34]

    Sumsub Research: Global Deepfake Incidents Surge Tenfold From 2022 to 2023 [Internet]

    Sumsub. Sumsub Research: Global Deepfake Incidents Surge Tenfold From 2022 to 2023 [Internet]. Sumsub; 2023 Nov 28 [accessed 2025 Jan 9]. Available from: https://sumsub.com/blog/deepfake-incidents-surge-tenfold/

    work page 2022

  4. [37]

    Metade dos golpes cibernéticos usarão vozes sintéticas em 2025 [Internet]

    Revista Segurança Eletrônica (Redação). Metade dos golpes cibernéticos usarão vozes sintéticas em 2025 [Internet]. Revista Segurança Eletrônica; 2025 [accessed 2025 Jan 9]. Available from: https://www.revistasegurancaeletronica.com.br/noticias/metade-dos-golpes-ciberneticos- usarao-vozes-sinteticas-em-2025/

    work page 2025

  5. [1]

    The Evolution of Zero Trust Architecture (ZTA) from Concept to Implementation

    Nasiruzzaman M, Ali M, Salam I, Miraz MH. The Evolution of Zero Trust Architecture (ZTA) from Concept to Implementation. In: 29th International Conference on Information Technology (IT); 2025 Feb 19–22; Žabljak, Montenegro. p. 1–8. DOI: 10.1109/IT64745.2025.10930254

    work page arXiv 2025

  6. [2]

    Build Security Into Your Network’s DNA: The Zero Trust Network Architecture [Internet]

    Kindervag J. Build Security Into Your Network’s DNA: The Zero Trust Network Architecture [Internet]. Forrester Research; 2010 [accessed 2025 Jan 9]. Available from: https://www.forrester.com/report/Build-Security-Into-Your-Networks-DNA-The-Zero-Trust- Network-Architecture/RES61059

    work page 2010

  7. [3]

    Zero Trust Architecture

    Rose S, Borchert O, Mitchell S, Connelly S. Zero Trust Architecture. NIST Special Publication 800-207. Gaithersburg (MD): National Institute of Standards and Technology; 2020 Aug. Report No.: NIST SP 800-207. DOI: 10.6028/NIST.SP .800-207

    doi:10.6028/nist.sp 2020

  8. [4]

    Securing GenAI Multi-Agent Systems Against Tool Squatting: A Zero Trust Registry-Based Approach [Preprint]

    Narajala VS, Huang K, Habler I. Securing GenAI Multi-Agent Systems Against Tool Squatting: A Zero Trust Registry-Based Approach [Preprint]. arXiv:2504.19951 [cs.CR]; 2025 Apr 28 [accessed 2025 May 27]. Available from: https://arxiv.org/abs/2504.19951

    arXiv 2025

Show all 40 references
  1. [5]

    Proof-of-Social-Capital: Privacy-Preserving Consensus Protocol Replacing Stake for Social Capital [Preprint]

    Mariani J, Homoliak I. Proof-of-Social-Capital: Privacy-Preserving Consensus Protocol Replacing Stake for Social Capital [Preprint]. arXiv:2505.12144 [cs.CR]; 2025 May 27 [accessed 2025 May 27]. Available from: https://arxiv.org/abs/2505.12144

    2025 arXiv

  2. [6]

    Verifiable Credentials Data Model 1.0: Expressing verifiable information on the Web [Internet]

    World Wide Web Consortium (W3C). Verifiable Credentials Data Model 1.0: Expressing verifiable information on the Web [Internet]. W3C Recommendation; 2019 Nov 19 [accessed 2025 Jan 9]. Available from: https://www.w3.org/TR/vc-data-model/

    2019

  3. [7]

    Never trust, always verify

    Buck C, Olenberger C, Schweizer A, Völter F, Eymann T. “Never trust, always verify”: A multivocal literature review on current knowledge and research gaps of zero-trust. Computers & Security. 2021;110:102436. DOI: 10.1016/j.cose.2021.102436

    2021

  4. [8]

    Influence of awareness and training on cybersecurity

    McCrohan KF, Engel K, Harvey JW . Influence of awareness and training on cybersecurity. J Internet Commer. 2010;9(1):23–41. DOI: 10.1080/15332861.2010.487415

    2010

  5. [9]

    Cybersecurity Framework 2.0 (Draft) [Internet]

    National Institute of Standards and Technology (NIST). Cybersecurity Framework 2.0 (Draft) [Internet]. Gaithersburg (MD): NIST; 2023 [accessed 2025 Jan 9]. Available from: https://www.nist.gov/cyberframework/newframework 12

    2023

  6. [10]

    Cybersecurity awareness training programs: a cost–benefit analysis framework

    Zhang Z, He W, Li W, Abdous MH. Cybersecurity awareness training programs: a cost–benefit analysis framework. Ind Manage Data Syst. 2021;121(3):613–36. DOI: 10.1108/IMDS-04-2020- 0228

    2021 doi

  7. [11]

    APFed: Anti-poisoning attacks in privacy-preserving heterogeneous federated learning

    Chen X, Yu H, Jia X, Yu X. APFed: Anti-poisoning attacks in privacy-preserving heterogeneous federated learning. IEEE Trans Inf Forensics Secur. 2023;18:5749–61. DOI: 10.1109/TIFS.2023.3291623

    2023

  8. [12]

    ShieldFL: Mitigating model poisoning attacks in privacy- preserving federated learning

    Ma Z, Ma J, Miao Y , Li Y , Deng RH. ShieldFL: Mitigating model poisoning attacks in privacy- preserving federated learning. IEEE Trans Inf Forensics Secur. 2022;17:1639–54. DOI: 10.1109/TIFS.2021.3138854

    2022

  9. [13]

    FLDetector: Defending federated learning against model poisoning attacks via detecting malicious clients

    Zhang Z, Cao X, Jia J, Gong NZ. FLDetector: Defending federated learning against model poisoning attacks via detecting malicious clients. In: Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’22); 2022. p. 2545–55. DOI: 10.1145/3534678.3539455

    2022

  10. [14]

    Perception poisoning attacks in federated learning

    Chow KH, Liu L. Perception poisoning attacks in federated learning. In: Proceedings of the IEEE TPS-ISA; 2021. p. 146–55. DOI: 10.1109/TPS-ISA53648.2021.00031

    2021

  11. [15]

    Membership inference attack and defense for wireless signal classifiers with deep learning

    Shi Y , Sagduyu YE. Membership inference attack and defense for wireless signal classifiers with deep learning. IEEE Trans Mobile Comput. 2023;22(7):4032–43. DOI: 10.1109/TMC.2021.3132536

    2023

  12. [16]

    FLAME: taming backdoors in federated learning

    Nguyen TD, Rieger P , Chen H, Yalame H, Möllering H, Fereidooni H, et al. FLAME: taming backdoors in federated learning. In: Proceedings of the USENIX Security Symposium 2022; 2022. p. 1415–32. Available from: https://www.usenix.org/conference/usenixsecurity22/presentation/ngu...

    2022

  13. [17]

    FLTrust: Byzantine-robust federated learning via trust bootstrapping

    Cao X, Fang M, Liu J, Gong NZ. FLTrust: Byzantine-robust federated learning via trust bootstrapping. In: Proceedings of the NDSS Symposium 2021; 2021. (The Internet Society). Available from: https://www.ndss-symposium.org/ndss-paper/fltrust-byzantine-robust-federated- learning...

    2021

  14. [18]

    Verify and trust: A multidimensional survey of zero-trust security in the age of IoT

    Azad MA, Abdullah S, Arshad J, Lallie HS, Ahmed Y . Verify and trust: A multidimensional survey of zero-trust security in the age of IoT. Internet of Things. 2024;27:101227. DOI: 10.1016/j.iot.2022.101227

    2024

  15. [19]

    Trust aware continuous authorization for zero trust in consumer Internet of Things

    Dimitrakos T, Dilshener T, Kravtsov A, La Marra A, Martinelli F, Rizos A, et al. Trust aware continuous authorization for zero trust in consumer Internet of Things. In: Proceedings of the 19th IEEE International Conference on Trust, Security and Privacy in Computing and Commun...

    2020

  16. [20]

    A Novel Zero-Trust Identity Framework for Agentic AI: Decentralized Authentication and Fine-Grained Access Control [Preprint]

    Huang K, Narajala VS, Yeoh J, Ross J, Lambe M, Raskar R, et al. A Novel Zero-Trust Identity Framework for Agentic AI: Decentralized Authentication and Fine-Grained Access Control [Preprint]. arXiv:2505.19301 [cs.CR]; 2025 May 24 [accessed 2025 May 24]. Available from: https://...

    2025 arXiv

  17. [21]

    A novel zero-trust network access control scheme based on the security profile of devices and users

    García-Teodoro P , Camacho J, Maciá-Fernández G, Gómez-Hernández JA, López-Marín VJ. A novel zero-trust network access control scheme based on the security profile of devices and users. Comput Networks. 2022;212:109068. DOI: 10.1016/j.comnet.2022.109068

    2022

  18. [22]

    Decentralized Identifiers (DIDs) v1.0 [Internet]

    World Wide Web Consortium (W3C). Decentralized Identifiers (DIDs) v1.0 [Internet]. W3C Recommendation; 2022 Jul 19 [accessed 2025 Jan 9]. Available from: https://www.w3.org/TR/did- core/

    2022

  19. [23]

    Zero-Trust Foundation Models for Secure and Trustworthy AI in IoT: Threats, Principles, and Future Directions [Preprint]

    Li K, Dressler F, Li C, Yuan X, Li S, Ni W, et al. Zero-Trust Foundation Models for Secure and Trustworthy AI in IoT: Threats, Principles, and Future Directions [Preprint]. arXiv:2505.23792 [cs.CR]; 2025 May 28 [accessed 2025 May 28]. Available from: https://arxiv.org/abs/2505.23792

    2025 arXiv

  20. [24]

    To Trust Or Not To Trust Your Vision-Language Model’s Prediction [Preprint]

    Dong H, Liu M, Liang J, Chatzi E, Fink O. To Trust Or Not To Trust Your Vision-Language Model’s Prediction [Preprint]. arXiv:2505.23745 [cs.CV]; 2025 May 29 [accessed 2025 May 26]. Available from: https://arxiv.org/abs/2505.23745v1 13

    2025

  21. [25]

    Securing smart UAV delivery systems using zero trust principle-driven blockchain architecture

    Dong C, Pal S, An Q, Yao A, Jiang F, Li J, et al. Securing smart UAV delivery systems using zero trust principle-driven blockchain architecture. In: 2023 IEEE International Conference on Blockchain (Blockchain); 2023. p. 315–22. DOI: 10.1109/Blockchain56703.2023.00058

    2023

  22. [26]

    Application of data collected by endpoint detection and response systems for implementation of a network security system based on zero trust principles and the EigenTrust algorithm

    Kumar N, Kasbekar GS, Manjunath D. Application of data collected by endpoint detection and response systems for implementation of a network security system based on zero trust principles and the EigenTrust algorithm. ACM SIGMETRICS Perform Eval Rev. 2023;50(4):5–7. DOI: 10.114...

    2023

  23. [27]

    ARGANIDS: a novel network intrusion detection system based on adversarially regularized graph autoencoder

    Venturi A, Ferrari M, Marchetti M, Colajanni M. ARGANIDS: a novel network intrusion detection system based on adversarially regularized graph autoencoder. In: Proceedings of the 38th ACM/SIGAPP Symposium on Applied Computing (SAC ’23); 2023. p. 1540–8. DOI: 10.1145/3555776.3577745

    2023

  24. [28]

    Something Just Like TRuST

    Atil B, Sureddy N, Passonneau RJ. “Something Just Like TRuST”: Toxicity Recognition of Span and Target* [Preprint]. arXiv:2506.02326 [cs.CL]; 2025 Jun 10 [accessed 2025 Jun 10]. Available from: https://arxiv.org/abs/2506.02326

    2025

  25. [29]

    Zero-Trust Mobility-Aware Authentication Framework for Secure Vehicular Fog Computing Networks [Preprint]

    Ahmad T. Zero-Trust Mobility-Aware Authentication Framework for Secure Vehicular Fog Computing Networks [Preprint]. arXiv:2506.05355 [cs.NI]; 2025 Jun 14 [accessed 2025 Jun 14]. Available from: https://arxiv.org/abs/2506.05355v1

    2025 arXiv

  26. [30]

    Zero Trust Cybersecurity: Procedures and Considerations in Context [Preprint]

    Lund BD, Lee TH, Wang Z, Wang T, Mannuru NR. Zero Trust Cybersecurity: Procedures and Considerations in Context [Preprint]. (arXiv:2505.18872v1) 2025 [accessed 2025 Jun 30]. Unpublished manuscript

    2025 arXiv

  27. [31]

    LLMs Are Not Yet Ready for Deepfake Image Detection [Preprint]

    Tariq S, Nguyen D, Chamikara MAP , Wu T, Abuadbba A, Moore K. LLMs Are Not Yet Ready for Deepfake Image Detection [Preprint]. arXiv:2506.10474 [cs.CV]; 2025 Jun 19 [accessed 2025 Jun 19]. Available from: https://arxiv.org/abs/2506.10474

    2025 arXiv

  28. [35]

    Deepfake Generation and Detection: A Benchmark and Survey [Preprint]

    Pei G, Zhang J, Hu M, Zhang Z, Wang C, Wu Y , et al. Deepfake Generation and Detection: A Benchmark and Survey [Preprint]. arXiv:2403.17881 [cs.CV]; 2024 May 31 [accessed 2025 Jan 9]. Available from: https://arxiv.org/abs/2403.17881

    2024

  29. [36]

    2024 Deepfakes Guide and Statistics: 70% Aren’t Confident They Can Spot a Deepfake [Internet]

    Security.org (Walsh M). 2024 Deepfakes Guide and Statistics: 70% Aren’t Confident They Can Spot a Deepfake [Internet]. Security.org; 2024 Feb 2 [accessed 2025 Jun 30]. Available from: https://www.security.org/resources/deepfake-statistics/

    2024

  30. [38]

    Out-of-Band Authentication: An Overview of Alternate Verification Channels [Internet]

    KZero. Out-of-Band Authentication: An Overview of Alternate Verification Channels [Internet]. KZero; 2024 [accessed 2025 Jan 9]. Available from: https://kzero.com/out-of-band-authentication- an-overview-of-alternate-verification-channels/

    2024

  31. [39]

    Deepfake: 66% dos brasileiros não sabem do que se trata [Internet]

    Embratel (Próximo Nível). Deepfake: 66% dos brasileiros não sabem do que se trata [Internet]. Próximo Nível – Embratel; 2024 May 6 [accessed 2025 Jun 30]. Available from: https://proximonivel.embratel.com.br/mais-de-60-dos-brasileiros-nao-sabem-o-que-e-deepfake/

    2024

  32. [40]

    Golpistas usam deepfake de diretor financeiro e roubam US$ 25 milhões [Internet]

    CNN Brasil. Golpistas usam deepfake de diretor financeiro e roubam US$ 25 milhões [Internet]. CNN Brasil; 2024 [accessed 2025 Jan 9]. Available from: 14 https://www.cnnbrasil.com.br/business/golpistas-usam-deepfake-de-diretor-financeiro-e- roubam-us-25-milhoes/

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.