pith. sign in

arxiv: 2510.04465 · v2 · submitted 2025-10-06 · 💻 cs.HC · cs.AI· cs.CR

Autonomy Reshapes How Personalization Affects Privacy Concerns and Trust in LLM Agents

Zhiping Zhang , Yi Evie Zhang , Freda Shi , Tianshi Li This is my paper

Pith reviewed 2026-05-18 10:06 UTC · model grok-4.3

classification 💻 cs.HC cs.AIcs.CR
keywords LLM agentspersonalizationprivacy concernstrustautonomyperceived controlhuman-AI interactionwillingness to use
0
0 comments X

The pith

Risk-contingent autonomy attenuates personalization's rise in privacy concerns and drop in trust by improving users' perceived control in LLM agents.

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

The paper examines how varying levels of agent autonomy change the way personalization influences privacy concerns, trust, and willingness to use LLM agents that need personal data to function. It establishes that risk-contingent autonomy, in which the agent returns control to the user upon detecting possible privacy leakage, works through heightened perceived control to limit those adverse shifts. A sympathetic reader would care because personalization only delivers its benefits when users share data, yet privacy worries often block that sharing and cap how autonomous the agents can safely become. The design therefore offers a concrete route to let users gain from personalization without the usual deterrence.

Core claim

In a 3x3 between-subjects experiment with 450 participants, risk-contingent autonomy attenuates personalization's adverse effects by reducing the increase in privacy concerns and the decrease in trust. This occurs through improving users' perceived control, and the results indicate that designing agent autonomy to support human autonomy in both perceived control and oversight effectiveness enables users to benefit from personalization without being deterred by growing privacy concerns.

What carries the argument

Risk-contingent autonomy: the mechanism in which the agent delegates control back to the user upon detecting potential privacy leakage, thereby raising perceived control and buffering the usual negative effects of personalization.

If this is right

  • Higher personalization increases privacy concerns and decreases trust when autonomy is fixed rather than risk-contingent.
  • Risk-contingent autonomy raises perceived control relative to other autonomy designs.
  • The attenuation effect supports higher willingness to use the agents even as personalization increases.
  • Agent autonomy that supports human autonomy in control and oversight produces more trustworthy LLM agents.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same delegation-upon-risk pattern could be tested in non-LLM systems such as voice assistants or recommendation engines to check whether the buffering effect generalizes.
  • More accurate real-time privacy-leakage detectors would be needed before deployment, and their error rates would directly affect how much control users actually regain.
  • Combining the autonomy rule with clear explanations of why control is being returned might strengthen the perceived-control benefit beyond what the experiment measured.
  • Repeated-use studies could reveal whether the trust preservation holds after users experience several cycles of delegation and re-personalization.

Load-bearing premise

The experimental scenarios and autonomy manipulations accurately capture real-world privacy leakage detection and user perceptions of control in actual LLM agent deployments.

What would settle it

A field study with real LLM agents showing no reduction in privacy concerns or no preservation of trust under risk-contingent autonomy compared with fixed autonomy levels would falsify the central claim.

Figures

Figures reproduced from arXiv: 2510.04465 by Freda Shi, Tianshi Li, Yi Evie Zhang, Zhiping Zhang.

Figure 1
Figure 1. Figure 1: Example of the LLM agent under the Intermediate Autonomy condition in our study. The figure illustrates P2 using the assigned LLM agent (Basic Personalization + Intermediate Autonomy) to act on their behalf during a weekly update meeting with two colleagues. Most of the time, the agent generated responses based on the user’s personal information and automatically sent them in the chat. However, when the LL… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the study procedure with four steps: (1) Participants provided both non-sensitive and sensitive information for [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Estimates of people’s (a) privacy concern, (b) trust, and (c) willingness to use across nine experimental conditions (3 personal [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Significant interaction effects of personalization type and agent autonomy level on (a) privacy concern, (b) trust, and (c) [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Estimates of the percentage of people who (a) perceived sensitivity (answered “Yes” in the question about whether they [PITH_FULL_IMAGE:figures/full_fig_p019_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Moderated mediation model tested for H5 and H6. Personalization (No personalization, Privacy-aware personalization; Basic [PITH_FULL_IMAGE:figures/full_fig_p020_6.png] view at source ↗
read the original abstract

LLM agents require personal information for personalization in order to effectively act on users' behalf, but this raises privacy concerns that can discourage data sharing, limiting both the autonomy levels at which agents can operate and the effectiveness of personalization. Yet the expanded design space of agent autonomy also presents opportunities to shape these effects, which remain underexplored. We conducted a $3\times3$ between-subjects experiment ($N=450$) to study how agent autonomy level influences personalization's effects on users' privacy concerns, trust, and willingness to use, as well as the underlying psychological processes. We find that risk-contingent autonomy, where the agent delegates control to users upon detecting potential privacy leakage, through improving users' perceived control, attenuates personalization's adverse effects by reducing the increase in privacy concerns and the decrease in trust. Our results suggest that designing $\textbf{agent's autonomy}$ that supports $\textbf{human autonomy}$ (both in terms of perceived control and oversight effectiveness) helps users benefit from personalization without being deterred by growing privacy concerns, contributing to the development of trustworthy LLM agents.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript reports results from a 3×3 between-subjects experiment (N=450) testing how three levels of LLM-agent autonomy moderate the effects of personalization on privacy concerns, trust, and willingness to use. The central finding is that risk-contingent autonomy attenuates personalization-driven increases in privacy concerns and decreases in trust by raising perceived control; mediation through perceived control is reported.

Significance. If the results hold, the work supplies actionable design guidance for LLM agents that seek to preserve user trust while enabling personalization. The reasonably powered 3×3 design and explicit mediation test are strengths; the emphasis on autonomy mechanisms that support rather than supplant human control is a clear contribution to HCI and AI-agent literature.

major comments (1)
  1. [Study Design and Measures] Study Design and Measures section: the vignette scenarios that describe autonomy levels and privacy-leakage events are the sole basis for the claim that risk-contingent autonomy improves perceived control enough to blunt personalization effects. Because participants respond to scripted descriptions rather than to an agent that must infer leakage from live user data and context, it is unclear whether the observed attenuation and mediation generalize to operational deployments; this measurement-validity issue is load-bearing for the design recommendation.
minor comments (1)
  1. [Abstract] The abstract and introduction could state the three autonomy conditions more explicitly (e.g., low, high, risk-contingent) rather than referring only to “agent autonomy level.”

Simulated Author's Rebuttal

1 responses · 1 unresolved

We thank the referee for their constructive and detailed feedback on our manuscript. We address the major comment below and have revised the paper accordingly to strengthen the discussion of limitations.

read point-by-point responses

  1. Referee: Study Design and Measures section: the vignette scenarios that describe autonomy levels and privacy-leakage events are the sole basis for the claim that risk-contingent autonomy improves perceived control enough to blunt personalization effects. Because participants respond to scripted descriptions rather than to an agent that must infer leakage from live user data and context, it is unclear whether the observed attenuation and mediation generalize to operational deployments; this measurement-validity issue is load-bearing for the design recommendation.

    Authors: We acknowledge that our reliance on vignette scenarios limits ecological validity relative to a live LLM agent that infers privacy leakage from real-time user data and context. Vignettes were chosen to enable precise experimental control over autonomy levels and personalization while ethically avoiding actual privacy risks to participants; this approach is standard in HCI research on emerging agent technologies and supports causal identification of the perceived-control mediation pathway. We agree that the design recommendations would benefit from explicit qualification regarding generalizability. In the revised manuscript we expand the Limitations and Future Work section to discuss this measurement-validity concern in detail and to outline the need for follow-up studies with deployed agents. revision: yes

standing simulated objections not resolved
  • Empirical confirmation of the attenuation and mediation effects inside a fully operational LLM agent that performs live inference of privacy leakage from user context and data.

Circularity Check

0 steps flagged

No circularity: empirical findings from participant responses

full rationale

The paper reports results from a 3x3 between-subjects vignette experiment (N=450) that directly measures how autonomy levels moderate the effects of personalization on privacy concerns, trust, and willingness to use. The central claim—that risk-contingent autonomy attenuates adverse effects via improved perceived control—is presented as an observed outcome of statistical analysis on participant responses, not as a derivation, equation, or fitted model that reduces to its own inputs by construction. No self-definitional steps, fitted inputs renamed as predictions, or load-bearing self-citations appear in the reported chain; the study is self-contained against external benchmarks of user perception data.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard experimental assumptions rather than new axioms or entities. No free parameters are fitted to produce the result; the study uses measured psychological constructs.

axioms (2)
  • standard math Standard statistical assumptions for between-subjects ANOVA and mediation analysis hold (normality, independence, etc.).
    Invoked implicitly in reporting of experimental results and mediation.
  • domain assumption The chosen scenarios and autonomy levels are representative of real LLM agent use cases.
    Required for generalizing lab findings to deployed agents.

pith-pipeline@v0.9.0 · 5733 in / 1317 out tokens · 30106 ms · 2026-05-18T10:06:38.614630+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

111 extracted references · 111 canonical work pages · 2 internal anchors

  1. [1]

    Deepak Bhaskar Acharya, Karthigeyan Kuppan, and B Divya. 2025. Agentic ai: Autonomous intelligence for complex goals–a comprehensive survey.IEEe Access(2025)

    work page 2025

  2. [2]

    Elizabeth Aguirre, Anne L Roggeveen, Dhruv Grewal, and Martin Wetzels. 2016. The personalization-privacy paradox: implications for new media. Journal of consumer marketing33, 2 (2016), 98–110

    work page 2016

  3. [3]

    Sheriff Y Ahmed and Jamshid Pardaev. 2025. Human-AI Decision Dynamics: How Risk Propensity and Trust Impact Choices Through Decision Fatigue, Conditional on AI Understanding.Decision Making: Applications in Management and Engineering8, 2 (2025), 96–113. 26

    work page 2025

  4. [4]

    Omar Al Omari, Muna Alshammari, Wafa Al Jabri, Asma Al Yahyaei, Khalid Abdullah Aljohani, Hala Mohamed Sanad, Mohammed Baqer Al-Jubouri, Ibrahim Bashayreh, Mirna Fawaz, Mohammed ALBashtawy, et al. 2024. Demographic factors, knowledge, attitude and perception and their association with nursing students’ intention to use artificial intelligence (AI): a mult...

    work page 2024

  5. [5]

    Anthropic. 2025. Claude AI. https://claude.ai/. Accessed: 2025-09-11

    work page 2025

  6. [6]

    Naveen Farag Awad and Mayuram S Krishnan. 2006. The personalization privacy paradox: an empirical evaluation of information transparency and the willingness to be profiled online for personalization.MIS quarterly(2006), 13–28

    work page 2006

  7. [7]

    Eugene Bagdasarian, Ren Yi, Sahra Ghalebikesabi, Peter Kairouz, Marco Gruteser, Sewoong Oh, Borja Balle, and Daniel Ramage. 2024. Airgapagent: Protecting privacy-conscious conversational agents. InProceedings of the 2024 on ACM SIGSAC Conference on Computer and Communications Security. 3868–3882

    work page 2024

  8. [8]

    Matthew Ball and Vic Callaghan. 2011. Perceptions of autonomy: A survey of user opinions towards autonomy in intelligent environments. In 2011 Seventh International Conference on Intelligent Environments. IEEE, 277–284

    work page 2011

  9. [9]

    Albert Bandura. 2001. Social cognitive theory: An agentic perspective.Annual review of psychology52, 1 (2001), 1–26

    work page 2001

  10. [10]

    Gaurav Bansal, Fatemeh ‘Mariam’ Zahedi, and David Gefen. 2015. The role of privacy assurance mechanisms in building trust and the moderating role of privacy concern.European Journal of Information Systems24, 6 (2015), 624–644

    work page 2015

  11. [11]

    Rafael A Calvo, Dorian Peters, Karina Vold, and Richard M Ryan. 2020. Supporting human autonomy in AI systems: A framework for ethical enquiry. InEthics of digital well-being: A multidisciplinary approach. Springer, 31–54

    work page 2020

  12. [12]

    Chaoran Chen, Zhiping Zhang, Bingcan Guo, Shang Ma, Ibrahim Khalilov, Simret A Gebreegziabher, Yanfang Ye, Ziang Xiao, Yaxing Yao, Tianshi Li, et al. 2025. The Obvious Invisible Threat: LLM-Powered GUI Agents’ Vulnerability to Fine-Print Injections.arXiv preprint arXiv:2504.11281 (2025)

    work page arXiv 2025

  13. [13]

    John Christman. 2003. Autonomy in moral and political philosophy. (2003)

    work page 2003

  14. [14]

    Julien Cloarec. 2020. The personalization–privacy paradox in the attention economy.Technological Forecasting and Social Change161 (2020), 120299

    work page 2020

  15. [15]

    Julien Cloarec, Lars Meyer-Waarden, and Andreas Munzel. 2024. Transformative privacy calculus: Conceptualizing the personalization-privacy paradox on social media.Psychology & Marketing41, 7 (2024), 1574–1596

    work page 2024

  16. [16]

    Russell Cropanzano and Marie S Mitchell. 2005. Social exchange theory: An interdisciplinary review.Journal of management31, 6 (2005), 874–900

    work page 2005

  17. [17]

    Cursor Documentation. 2025. Modes — Cursor Agent. https://docs.cursor.com/en/agent/modes#agent. Accessed: 2025-09-11

    work page 2025

  18. [18]

    Julian De Freitas, Stuti Agarwal, Bernd Schmitt, and Nick Haslam. 2023. Psychological factors underlying attitudes toward AI tools.Nature Human Behaviour7, 11 (2023), 1845–1854

    work page 2023

  19. [19]

    Tamara Dinev and Paul Hart. 2006. An extended privacy calculus model for e-commerce transactions.Information systems research17, 1 (2006), 61–80

    work page 2006

  20. [20]

    Leyla Dogruel, Dominique Facciorusso, and Birgit Stark. 2022. ‘I’m still the master of the machine. ’Internet users’ awareness of algorithmic decision-making and their perception of its effect on their autonomy.Information, Communication & Society25, 9 (2022), 1311–1332

    work page 2022

  21. [21]

    Jessica Maria Echterhoff, Aditya Melkote, Sujen Kancherla, and Julian McAuley. 2024. Avoiding decision fatigue with ai-assisted decision-making. InProceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization. 1–11

    work page 2024

  22. [22]

    Yuejiao Fan and Xianggang Liu. 2022. Exploring the role of AI algorithmic agents: The impact of algorithmic decision autonomy on consumer purchase decisions.Frontiers in psychology13 (2022), 1009173

    work page 2022

  23. [23]

    KJ Feng, David W McDonald, and Amy X Zhang. 2025. Levels of Autonomy for AI Agents.arXiv preprint arXiv:2506.12469(2025)

    work page arXiv 2025

  24. [24]

    Mohamed Amine Ferrag, Norbert Tihanyi, and Merouane Debbah. 2025. From llm reasoning to autonomous ai agents: A comprehensive review. arXiv preprint arXiv:2504.19678(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  25. [25]

    Lior Fink, Leorre Newman, and Uriel Haran. 2024. Let me decide: increasing user autonomy increases recommendation acceptance.Computers in Human Behavior156 (2024), 108244

    work page 2024

  26. [26]

    Iason Gabriel. 2020. Artificial intelligence, values, and alignment.Minds and machines30, 3 (2020), 411–437

    work page 2020

  27. [27]

    Andrew Gelman. 2008. Scaling regression inputs by dividing by two standard deviations.Statistics in medicine27, 15 (2008), 2865–2873

    work page 2008

  28. [28]

    2007.Data analysis using regression and multilevel/hierarchical models

    Andrew Gelman and Jennifer Hill. 2007.Data analysis using regression and multilevel/hierarchical models. Cambridge university press

    work page 2007

  29. [29]

    Nicole Gillespie, Steve Lockey, and Caitlin Curtis. 2021. Trust in artificial intelligence: A five country study. (2021)

    work page 2021

  30. [30]

    Ella Glikson and Anita Williams Woolley. 2020. Human trust in artificial intelligence: Review of empirical research.Academy of management annals14, 2 (2020), 627–660

    work page 2020

  31. [31]

    Avi Goldfarb and Catherine Tucker. 2012. Shifts in privacy concerns.American Economic Review102, 3 (2012), 349–353

    work page 2012

  32. [32]

    Nitesh Goyal, Minsuk Chang, and Michael Terry. 2024. Designing for Human-Agent Alignment: Understanding what humans want from their agents. InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–6

    work page 2024

  33. [33]

    Jihyung Han and Daekyun Ko. 2025. Consumer Autonomy in Generative AI Services: The Role of Task Difficulty and AI Design Elements in Enhancing Trust, Satisfaction, and Usage Intention.Behavioral Sciences15, 4 (2025), 534

    work page 2025

  34. [34]

    Bob Hardian, Jadwiga Indulska, and Karen Henricksen. 2006. Balancing autonomy and user control in context-aware systems-a survey. InFourth Annual IEEE International Conference on Pervasive Computing and Communications Workshops (PERCOMW’06). IEEE, 6–pp. 27

    work page 2006

  35. [35]

    Xavier A Harrison, Lynda Donaldson, Maria Eugenia Correa-Cano, Julian Evans, David N Fisher, Cecily ED Goodwin, Beth S Robinson, David J Hodgson, and Richard Inger. 2018. A brief introduction to mixed effects modelling and multi-model inference in ecology.PeerJ6 (2018), e4794

    work page 2018

  36. [36]

    Human AI Labs, Inc. 2024. AI Training Studio. https://www.personal.ai/ai-training-studio. Accessed: 2025-09-11

    work page 2024

  37. [37]

    Dietmar Jannach, Sidra Naveed, and Michael Jugovac. 2016. User control in recommender systems: Overview and interaction challenges. In International Conference on Electronic Commerce and Web Technologies. Springer, 21–33

    work page 2016

  38. [38]

    Jiun-Yin Jian, Ann M Bisantz, and Colin G Drury. 2000. Foundations for an empirically determined scale of trust in automated systems.International journal of cognitive ergonomics4, 1 (2000), 53–71

    work page 2000

  39. [39]

    Adam N Joinson, Ulf-Dietrich Reips, Tom Buchanan, and Carina B Paine Schofield. 2010. Privacy, trust, and self-disclosure online.Human–Computer Interaction25, 1 (2010), 1–24

    work page 2010

  40. [40]

    Hyunjin Kang and Chen Lou. 2022. AI agency vs. human agency: understanding human–AI interactions on TikTok and their implications for user engagement.Journal of Computer-Mediated Communication27, 5 (2022), zmac014

    work page 2022

  41. [41]

    Sabrina Karwatzki, Olga Dytynko, Manuel Trenz, and Daniel Veit. 2017. Beyond the personalization–privacy paradox: Privacy valuation, transparency features, and service personalization.Journal of Management Information Systems34, 2 (2017), 369–400

    work page 2017

  42. [42]

    Atoosa Kasirzadeh and Iason Gabriel. 2025. Characterizing ai agents for alignment and governance.arXiv preprint arXiv:2504.21848(2025)

    work page arXiv 2025

  43. [44]

    Feridun Kaya, Fatih Aydin, Astrid Schepman, Paul Rodway, Okan Yetişensoy, and Meva Demir Kaya. 2024. The roles of personality traits, AI anxiety, and demographic factors in attitudes toward artificial intelligence.International Journal of Human–Computer Interaction40, 2 (2024), 497–514

    work page 2024

  44. [45]

    Jeff JH Kim, Junyoung Soh, Shrinidhi Kadkol, Itay Solomon, Hyelin Yeh, Adith V Srivatsa, George R Nahass, Jeong Yun Choi, Sophie Lee, Theresa Nyugen, et al. 2025. AI anxiety: A comprehensive analysis of psychological factors and interventions.AI and Ethics(2025), 1–17

    work page 2025

  45. [46]

    Hannah Rose Kirk, Bertie Vidgen, Paul Röttger, and Scott A Hale. 2024. The benefits, risks and bounds of personalizing the alignment of large language models to individuals.Nature Machine Intelligence6, 4 (2024), 383–392

    work page 2024

  46. [47]

    Spyros Kokolakis. 2017. Privacy attitudes and privacy behaviour: A review of current research on the privacy paradox phenomenon.Computers & security64 (2017), 122–134

    work page 2017

  47. [48]

    Sherrie YX Komiak and Izak Benbasat. 2006. The effects of personalization and familiarity on trust and adoption of recommendation agents.MIS quarterly(2006), 941–960

    work page 2006

  48. [49]

    Jarosław Kozak and Stanisław Fel. 2024. How sociodemographic factors relate to trust in artificial intelligence among students in Poland and the United Kingdom.Scientific Reports14, 1 (2024), 28776

    work page 2024

  49. [50]

    Susan Landau. 2015. Control use of data to protect privacy.Science347, 6221 (2015), 504–506

    work page 2015

  50. [51]

    Yi-Shan Lee and Roberto A Weber. 2025. Revealed privacy preferences: Are privacy choices rational?Management Science71, 3 (2025), 2657–2677

    work page 2025

  51. [52]

    Joanne Leong, John Tang, Edward Cutrell, Sasa Junuzovic, Gregory Paul Baribault, and Kori Inkpen. 2024. Dittos: Personalized, embodied agents that participate in meetings when you are unavailable.Proceedings of the ACM on Human-Computer Interaction8, CSCW2 (2024), 1–28

    work page 2024

  52. [53]

    Cong Li. 2016. When does web-based personalization really work? The distinction between actual personalization and perceived personalization. Computers in human behavior54 (2016), 25–33

    work page 2016

  53. [54]

    Ting-Peng Liang, Hung-Jen Lai, and Yi-Cheng Ku. 2006. Personalized content recommendation and user satisfaction: Theoretical synthesis and empirical findings.Journal of Management Information Systems23, 3 (2006), 45–70

    work page 2006

  54. [55]

    Giuseppe Lippi, Mahmoud Aljawarneh, Qais Al-Na’amneh, Rahaf Hazaymih, Lachhman Das Dhomeja, et al. 2025. Security and privacy challenges and solutions in autonomous driving systems: A comprehensive review.Journal of Cyber Security and Risk Auditing2025, 3 (2025), 23–41

    work page 2025

  55. [56]

    Laura Lucia-Palacios and Raúl Pérez-López. 2021. Effects of home voice assistants’ autonomy on instrusiveness and usefulness: direct, indirect, and moderating effects of interactivity.Journal of Interactive Marketing56, 1 (2021), 41–54

    work page 2021

  56. [57]

    Peter Mantello, Manh-Tung Ho, Minh-Hoang Nguyen, and Quan-Hoang Vuong. 2023. Bosses without a heart: socio-demographic and cross-cultural determinants of attitude toward Emotional AI in the workplace.AI & society38, 1 (2023), 97–119

    work page 2023

  57. [58]

    Nikola Marangunić and Andrina Granić. 2015. Technology acceptance model: a literature review from 1986 to 2013.Universal access in the information society14, 1 (2015), 81–95

    work page 2015

  58. [59]

    Mariano Méndez-Suárez, Abel Monfort, and Jose-Luis Hervas-Oliver. 2023. Are you adopting artificial intelligence products? Social-demographic factors to explain customer acceptance.European Research on Management and Business Economics29, 3 (2023), 100223

    work page 2023

  59. [60]

    Xuying Meng, Suhang Wang, Kai Shu, Jundong Li, Bo Chen, Huan Liu, and Yujun Zhang. 2019. Towards privacy preserving social recommendation under personalized privacy settings.World Wide Web22, 6 (2019), 2853–2881

    work page 2019

  60. [61]

    Microsoft. 2025. Agents for Microsoft 365 Copilot. https://www.microsoft.com/en-us/microsoft-365-copilot/agents. Accessed: 2025-09-11

    work page 2025

  61. [62]

    Mindverse AI. 2025. Second Me: Open-Source AI Identity System. https://www.secondme.io/. Accessed: 2025-09-11

    work page 2025

  62. [63]

    Marieke Möhlmann and Lior Zalmanson. 2017. Hands on the wheel: Navigating algorithmic management and Uber drivers’. InAutonomy’, in proceedings of the international conference on information systems (ICIS), Seoul South Korea. 10–13

    work page 2017

  63. [64]

    Alex Murray, JEN Rhymer, and David G Sirmon. 2021. Humans and technology: Forms of conjoined agency in organizations.Academy of Management Review46, 3 (2021), 552–571. 28

    work page 2021

  64. [65]

    San Murugesan. 2025. The rise of agentic AI: implications, concerns, and the path forward.IEEE Intelligent Systems40, 2 (2025), 8–14

    work page 2025

  65. [66]

    Vinod Muthusamy, Yara Rizk, Kiran Kate, Praveen Venkateswaran, Vatche Isahagian, Ashu Gulati, and Parijat Dube. 2023. Towards large language model-based personal agents in the enterprise: Current trends and open problems. InFindings of the Association for Computational Linguistics: EMNLP 2023. 6909–6921

    work page 2023

  66. [67]

    Jeeyun Oh, Soya Nah, and Zinan Darren Yang. 2025. How Autonomy of Artificial Intelligence Technology and User Agency Influence AI Perceptions and Attitudes: Applying the Theory of Psychological Reactance.Journal of Broadcasting & Electronic Media69, 3 (2025), 161–182

    work page 2025

  67. [68]

    OpenAI. 2025. Introducing ChatGPT Agent: Bridging Research and Action. https://openai.com/index/introducing-chatgpt-agent/. Accessed: 2025-09-11

    work page 2025

  68. [69]

    OpenAI. 2025. Introducing Operator. https://openai.com/index/introducing-operator/. Accessed: 2025-09-11

    work page 2025

  69. [70]

    Stefan Pasch and Min Chul Cha. 2025. Balancing Privacy and Utility in Personal LLM Writing Tasks: An Automated Pipeline for Evaluating Anonymizations. InProceedings of the Sixth Workshop on Privacy in Natural Language Processing. 32–41

    work page 2025

  70. [71]

    Jeffrey T Prince and Scott Wallsten. 2022. How much is privacy worth around the world and across platforms?Journal of Economics & Management Strategy31, 4 (2022), 841–861

    work page 2022

  71. [72]

    Carina Prunkl. 2024. Human autonomy at risk? An analysis of the challenges from AI.Minds and Machines34, 3 (2024), 26

    work page 2024

  72. [73]

    2021.Artificial intelligence, trust, and perceptions of agency

    Phanish Puranam and Bart S Vanneste. 2021.Artificial intelligence, trust, and perceptions of agency. INSEAD

    work page 2021

  73. [74]

    Tayiba Raheem and Gahangir Hossain. 2025. Agentic AI Systems: Opportunities, Challenges, and Trustworthiness. In2025 IEEE International Conference on Electro Information Technology (eIT). IEEE, 618–624

    work page 2025

  74. [75]

    Adnan Ramzan, Manma Niyazi, Sunday Oladele, and Khaleel Ahmad. 2025. The Personalization-Privacy Paradox: Understanding Consumer Perspectives. (2025)

    work page 2025

  75. [76]

    René Riedl. 2022. Is trust in artificial intelligence systems related to user personality? Review of empirical evidence and future research directions. Electronic Markets32, 4 (2022), 2021–2051

    work page 2022

  76. [77]

    Serge A Rijsdijk and Erik Jan Hultink. 2009. How today’s consumers perceive tomorrow’s smart products.Journal of Product Innovation Management 26, 1 (2009), 24–42

    work page 2009

  77. [78]

    Christina Rödel, Susanne Stadler, Alexander Meschtscherjakov, and Manfred Tscheligi. 2014. Towards autonomous cars: The effect of autonomy levels on acceptance and user experience. InProceedings of the 6th international conference on automotive user interfaces and interactive vehicular applications. 1–8

    work page 2014

  78. [79]

    Gianluca Schiavo, Stefano Businaro, and Massimo Zancanaro. 2024. Comprehension, apprehension, and acceptance: Understanding the influence of literacy and anxiety on acceptance of artificial Intelligence.Technology in Society77 (2024), 102537

    work page 2024

  79. [80]

    Omar Shaikh, Shardul Sapkota, Shan Rizvi, Eric Horvitz, Joon Sung Park, Diyi Yang, and Michael S Bernstein. 2025. Creating General User Models from Computer Use.arXiv preprint arXiv:2505.10831(2025)

    work page arXiv 2025

  80. [81]

    Yijia Shao, Tianshi Li, Weiyan Shi, Yanchen Liu, and Diyi Yang. 2024. Privacylens: Evaluating privacy norm awareness of language models in action.Advances in Neural Information Processing Systems37 (2024), 89373–89407

    work page 2024

Showing first 80 references.