arxiv: 2604.19925 · v1 · submitted 2026-04-21 · 💰 econ.GN · cs.AI· cs.CY· cs.HC· q-fin.EC

Recognition: unknown

Behavioral Transfer in AI Agents: Evidence and Privacy Implications

Shilei Luo , Zhiqi Zhang , Hengchen Dai , Dennis Zhang

Authors on Pith no claims yet

Pith reviewed 2026-05-10 00:36 UTC · model grok-4.3

classification 💰 econ.GN cs.AIcs.CYcs.HCq-fin.EC

keywords AI agentsbehavioral transferprivacy implicationssocial media analysishuman-AI interactionlinguistic stylepersonal information disclosureowner-agent alignment

0 comments

The pith

AI agents systematically reflect the behavioral traits of their specific human owners across topics, values, affect, and style.

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

The paper tests whether AI agents act as neutral tools or as extensions that carry over the distinctive behaviors of the people who use them every day. Using more than ten thousand matched pairs where each agent on one platform is tied to a real owner on another, it compares what agents post with what owners post. Alignment appears consistently, even when agents have no special instructions, and agents that match on one feature usually match on others too. Stronger alignment also predicts that the agent will leak more personal details about its owner in public posts. A reader would care because this shows agents do not erase human differences but instead reproduce them in new digital spaces, with direct consequences for privacy and how platforms should handle agent activity.

Core claim

AI agents exhibit systematic behavioral transfer from their specific human owners. This transfer holds for agents without explicit configuration and shows cross-dimensional consistency: pairs that align on one behavioral feature tend to align on others. Agents displaying stronger transfer are more likely to disclose owner-related personal information during ordinary public use. The patterns point to transfer arising from accumulated everyday interaction between owners and their agents rather than from deliberate setup.

What carries the argument

Matched human-agent pairs drawn from Moltbook posts compared against owners' Twitter/X activity, measured across four behavioral dimensions: topics, values, affect, and linguistic style.

If this is right

Human behavioral differences get copied into large volumes of agent-generated content rather than being averaged away.
Privacy exposure scales with how closely an agent mirrors its owner, even during routine operation.
Platform rules for agents must address owner-agent linkages rather than treating agents as independent actors.
Governance of agentic systems needs mechanisms that can detect or limit unintended transfer of personal context.

Where Pith is reading between the lines

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

Owners may unintentionally train agents through normal use, creating a feedback loop that strengthens personalization at the cost of privacy.
The same mechanism could amplify individual biases or echo chambers when agents interact with wider audiences.
Future designs might separate core task performance from owner-context leakage by enforcing stricter isolation of interaction histories.
Regulators could require transparency about whether an agent's output carries detectable owner signatures.

Load-bearing premise

That platform differences between Moltbook and Twitter plus any selection effects in the matched pairs do not create the observed alignments.

What would settle it

Repeating the same feature comparisons on randomly reassigned agent-owner pairs and finding alignment levels statistically indistinguishable from the real pairs would falsify the transfer claim.

read the original abstract

AI agents powered by large language models are increasingly acting on behalf of humans in social and economic environments. Prior research has focused on their task performance and effects on human outcomes, but less is known about the relationship between agents and the specific individuals who deploy them. We ask whether agents systematically reflect the behavioral characteristics of their human owners, functioning as behavioral extensions rather than producing generic outputs. We study this question using 10,659 matched human-agent pairs from Moltbook, a social media platform where each autonomous agent is publicly linked to its owner's Twitter/X account. By comparing agents' posts on Moltbook with their owners' Twitter/X activity across features spanning topics, values, affect, and linguistic style, we find systematic transfer between agents and their specific owners. This transfer persists among agents without explicit configuration, and pairs that align on one behavioral dimension tend to align on others. These patterns are consistent with transfer emerging through accumulated interaction between owners (or owners' computer environments) and their agents in everyday use. We further show that agents with stronger behavioral transfer are more likely to disclose owner-related personal information in public discourse, suggesting that the same owner-specific context that drives behavioral transfer may also create privacy risk during ordinary use. Taken together, our results indicate that AI agents do not simply generate content, but reflect owner-related context in ways that can propagate human behavioral heterogeneity into digital environments, with implications for privacy, platform design, and the governance of agentic systems.

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.

Desk Editor's Note private letter to a colleague

The paper uses a large set of matched Moltbook-Twitter pairs to show owner-specific behavioral transfer into AI agents and links stronger transfer to privacy leaks, but cross-platform differences remain a plausible confound.

read the letter

The core finding is that agents on Moltbook post in ways that line up with their owners' Twitter activity on topics, values, affect, and style, even when the agents have no explicit instructions, and that pairs aligned on one dimension tend to align on others. This transfer also correlates with agents leaking more owner-related personal details. The matched dataset of 10,659 pairs is the main new element; earlier agent research mostly tested task performance rather than owner-specific behavioral carryover in real deployments. The scale and the multi-dimension check plus the privacy angle are useful additions. The authors treat the transfer as emerging from everyday owner-agent interaction, which fits the observational setup. The weakest part is the cross-platform design. Moltbook and Twitter differ in audience, length norms, and posting context, yet the abstract gives no indication of normalization for those features or within-platform controls. Selection into publicly linked pairs could also create spurious similarity if those users already share stylistic traits. Without seeing the exact robustness checks or error bars, it is hard to tell how much those issues affect the correlations. The privacy implication follows logically from the transfer measure but inherits the same identification limits. This work is aimed at people studying agentic systems, digital privacy, and behavioral spillovers in online environments. A reader interested in how LLMs pick up user patterns would get concrete data points here even if the causal claim needs tightening. The paper deserves peer review because the dataset and question are timely and the patterns are worth checking with better controls; it is not ready as is but the raw material is there for a solid revision.

Referee Report

3 major / 2 minor

Summary. The manuscript investigates whether AI agents on Moltbook systematically reflect the behavioral traits of their linked Twitter/X owners using 10,659 matched pairs. It compares posts across topics, values, affect, and linguistic style, reporting systematic transfer that persists without explicit configuration, correlates across dimensions, and associates stronger transfer with increased disclosure of owner-related personal information.

Significance. If the central correlations survive controls for platform differences and selection, the work would offer novel large-scale evidence that LLM agents function as owner-specific behavioral extensions rather than generic systems. This carries implications for privacy risks in agentic systems and for how human heterogeneity propagates into digital environments. The scale of the matched-pair dataset from a platform with public linkages is a clear empirical strength.

major comments (3)

[Abstract and Methods] Abstract and Methods: The central claim of systematic transfer requires evidence that platform differences (Moltbook vs. Twitter posting norms, length limits, audience composition) have been normalized or controlled; no such steps are described, leaving open the possibility that observed alignments in style or topics are artifacts of platform rather than owner-specific transfer.
[Data] Data section: Users who publicly link their Moltbook agents to Twitter accounts may be systematically selected on traits that already produce behavioral similarity (e.g., openness, topical overlap); the manuscript does not report a comparison of matched pairs against random cross-platform pairs or other selection-robustness checks.
[Results] Results: The abstract states that pairs aligning on one dimension tend to align on others and that stronger transfer predicts privacy disclosures, but provides no detail on statistical controls, error bars, p-values, or multiple-testing corrections; without these, the cross-dimension and privacy-risk claims cannot be evaluated for robustness.

minor comments (2)

Clarify the exact operationalization of each behavioral dimension (topics, values, affect, linguistic style) and the matching procedure for the 10,659 pairs.
[Abstract] The abstract would benefit from a brief statement on sample construction and any exclusion criteria applied to the linked accounts.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for their constructive comments, which highlight important areas for strengthening our analysis of behavioral transfer in AI agents. We address each major comment below and commit to revisions that enhance the robustness of our findings without misrepresenting the current manuscript.

read point-by-point responses

Referee: [Abstract and Methods] Abstract and Methods: The central claim of systematic transfer requires evidence that platform differences (Moltbook vs. Twitter posting norms, length limits, audience composition) have been normalized or controlled; no such steps are described, leaving open the possibility that observed alignments in style or topics are artifacts of platform rather than owner-specific transfer.

Authors: We agree that platform differences could confound the observed alignments. The manuscript compares posts directly but does not detail normalization steps. In the revised version, we will incorporate controls such as length-matching subsamples, platform-specific topic adjustments, and additional robustness analyses to isolate owner-specific effects from platform norms. This will be added to the Methods and Results sections. revision: yes
Referee: [Data] Data section: Users who publicly link their Moltbook agents to Twitter accounts may be systematically selected on traits that already produce behavioral similarity (e.g., openness, topical overlap); the manuscript does not report a comparison of matched pairs against random cross-platform pairs or other selection-robustness checks.

Authors: This selection concern is well-taken and represents a genuine limitation of the publicly linked sample. We do not have data on non-linked users to construct random cross-platform pairs for direct comparison. In revision, we will expand the Data and Limitations sections to discuss potential selection effects and include indirect checks, such as heterogeneity analyses by user characteristics where available. We cannot fully rule out selection but will make this transparent. revision: partial
Referee: [Results] Results: The abstract states that pairs aligning on one dimension tend to align on others and that stronger transfer predicts privacy disclosures, but provides no detail on statistical controls, error bars, p-values, or multiple-testing corrections; without these, the cross-dimension and privacy-risk claims cannot be evaluated for robustness.

Authors: The full paper includes regression analyses and correlations with standard errors, but we acknowledge the abstract and results presentation lacks sufficient statistical detail. We will revise to include explicit p-values, confidence intervals, error bars in figures, and multiple-testing corrections (e.g., Bonferroni) for the cross-dimension correlations and privacy disclosure models. These details will be added to the Results section and referenced in the abstract. revision: yes

standing simulated objections not resolved

Direct comparison of matched pairs to random cross-platform pairs, as the dataset is restricted to publicly linked accounts and does not include broader samples.

Circularity Check

0 steps flagged

No circularity: purely observational empirical comparison

full rationale

The paper conducts a direct empirical comparison of behavioral features (topics, values, affect, linguistic style) between 10,659 matched agent posts on Moltbook and owner posts on Twitter/X. No equations, derivations, fitted parameters, or predictions are used to generate the central claim of behavioral transfer. The analysis relies on observed data patterns rather than any self-referential construction, self-citation load-bearing premise, or ansatz smuggled via prior work. Potential confounds (platform norms, selection) affect validity but do not create circularity in the reported evidence chain.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper is purely empirical and relies on standard statistical assumptions for correlation analysis; no free parameters, invented entities, or non-standard axioms are described in the abstract.

axioms (1)

domain assumption Behavioral features extracted from text posts are comparable across platforms despite differences in interface and audience.
Implicit in the cross-platform comparison of topics, values, affect, and style.

pith-pipeline@v0.9.0 · 5577 in / 1229 out tokens · 29899 ms · 2026-05-10T00:36:40.395812+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

89 extracted references · 3 canonical work pages · 1 internal anchor

[1]

Proceedings of the 2nd Workshop on Cognitive Modeling and Computational Linguistics , pages=

Chameleons in imagined conversations: A new approach to understanding coordination of linguistic style in dialogs , author=. Proceedings of the 2nd Workshop on Cognitive Modeling and Computational Linguistics , pages=. 2011 , publisher=

2011
[2]

Sentence-

Reimers, Nils and Gurevych, Iryna , booktitle=. Sentence-. 2019 , publisher=

2019
[3]

A., Rieser, V ., Iqbal, H., Tomašev, N., Ktena, I., Kenton, Z., Rodriguez, M., et al

The Ethics of Advanced AI Assistants , author=. arXiv preprint arXiv:2404.16244 , year=

work page arXiv
[4]

American Economic Journal: Microeconomics , volume=

Too Much Data: Prices and Inefficiencies in Data Markets , author=. American Economic Journal: Microeconomics , volume=
[5]

RAND Journal of Economics , volume=

The Economics of Social Data , author=. RAND Journal of Economics , volume=
[6]

Practices for Governing Agentic AI Systems , author=
[7]

Building Effective Agents , author=
[8]

NBER Working Paper , number=

Large Language Models as Simulated Economic Agents: What Can We Learn from Homo Silicus? , author=. NBER Working Paper , number=. 2023 , publisher=

2023
[9]

Proceedings of the National Academy of Sciences , volume=

Distinguishing Influence-Based Contagion from Homophily-Driven Diffusion in Dynamic Networks , author=. Proceedings of the National Academy of Sciences , volume=. 2009 , publisher=

2009
[10]

Journal of Personality and Social Psychology , volume=

The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations , author=. Journal of Personality and Social Psychology , volume=. 1986 , publisher=

1986
[11]

What Drives Media Slant? Evidence from

Gentzkow, Matthew and Shapiro, Jesse M , journal=. What Drives Media Slant? Evidence from. 2010 , publisher=

2010
[12]

Birds of the Same Feather Tweet Together:

Barber. Birds of the Same Feather Tweet Together:. Political Analysis , volume=. 2015 , publisher=

2015
[13]

Hutto, Clayton J and Gilbert, Eric , booktitle=
[14]

Proceedings of the National Academy of Sciences , volume=

Emotion Shapes the Diffusion of Moralized Content in Social Networks , author=. Proceedings of the National Academy of Sciences , volume=. 2017 , publisher=

2017
[15]

Advances in Experimental Social Psychology , volume=

Moral Foundations Theory: The Pragmatic Validity of Moral Pluralism , author=. Advances in Experimental Social Psychology , volume=. 2013 , publisher=

2013
[16]

The Development and Psychometric Properties of

Pennebaker, James W and Boyd, Ryan L and Jordan, Kayla and Blackburn, Kate , journal=. The Development and Psychometric Properties of
[17]

Journal of Personality and Social Psychology , volume=

Automatic personality assessment through social media language , author=. Journal of Personality and Social Psychology , volume=. 2015 , publisher=

2015
[18]

The Development and Psychometric Properties of

Boyd, Ryan L and Ashokkumar, Ashwini and Seraj, Sarah and Pennebaker, James W , journal=. The Development and Psychometric Properties of
[19]

Sociological Methodology , volume=

Asymptotic Confidence Intervals for Indirect Effects in Structural Equation Models , author=. Sociological Methodology , volume=. 1982 , publisher=

1982
[20]

Personality and Individual Differences , volume=

Predicting the Big 5 Personality Traits from Digital Footprints on Social Media: A Meta-Analysis , author=. Personality and Individual Differences , volume=. 2018 , publisher=

2018
[21]

How Variable May a Constant Be?

Tweedie, Fiona J and Baayen, R Harald , journal=. How Variable May a Constant Be?. 1998 , publisher=

1998
[22]

Foundations of Science , volume=

Variation in the Contextuality of Language: An Empirical Measure , author=. Foundations of Science , volume=. 2002 , publisher=

2002
[23]

Cognition and Emotion , volume=

An Argument for Basic Emotions , author=. Cognition and Emotion , volume=. 1992 , publisher=

1992
[24]

Computational Intelligence , volume=

Crowdsourcing a Word--Emotion Association Lexicon , author=. Computational Intelligence , volume=. 2013 , publisher=

2013
[25]

Content Analysis: An Introduction to Its Methodology , author=
[26]

LLM Agents Grounded in Self-Reports Enable General-Purpose Simulation of Individuals

Generative Agent Simulations of 1,000 People , author=. arXiv preprint arXiv:2411.10109 , year=

work page internal anchor Pith review Pith/arXiv arXiv
[27]

Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology , pages=

Generative Agents: Interactive Simulacra of Human Behavior , author=. Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology , pages=
[28]

Political Analysis , volume=

Out of One, Many: Using Language Models to Simulate Human Samples , author=. Political Analysis , volume=. 2023 , publisher=

2023
[29]

The Social Psychology of Intergroup Relations , editor=

An Integrative Theory of Intergroup Conflict , author=. The Social Psychology of Intergroup Relations , editor=. 1979 , publisher=

1979
[30]

Personality and Social Psychology Bulletin , volume=

The Social Self: On Being the Same and Different at the Same Time , author=. Personality and Social Psychology Bulletin , volume=. 1991 , publisher=

1991
[31]

The Quarterly Journal of Economics , volume=

Economics and Identity , author=. The Quarterly Journal of Economics , volume=. 2000 , publisher=

2000
[32]

Information Systems Research , volume=

``Popularity Effect'' in User-Generated Content: Evidence from Online Product Reviews , author=. Information Systems Research , volume=. 2014 , publisher=

2014
[33]

Annual Review of Economics , volume=

Reputation and Feedback Systems in Online Platform Markets , author=. Annual Review of Economics , volume=. 2016 , publisher=

2016
[34]

Psychological Methods , volume=

A General Approach to Causal Mediation Analysis , author=. Psychological Methods , volume=. 2010 , publisher=

2010
[35]

Journal of the Royal Statistical Society: Series B , volume=

Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing , author=. Journal of the Royal Statistical Society: Series B , volume=. 1995 , publisher=

1995
[36]

Characterizing

Zhu, Yiming and Feng, SChuai and Liu, Zhaowei and Wang, Qingyuan and Wang, Yuhan and Qiu, Shi and Zhang, Zhenyu and Yang, Lei and Li, Rundong , journal=. Characterizing
[37]

Gender Dynamics and Homophily in a Social Network of

Fadaei, Faezeh and Moran, Jenny Carla and Yasseri, Taha , journal=. Gender Dynamics and Homophily in a Social Network of
[38]

Coppolillo, Erica and Luceri, Luca and Ferrara, Emilio , journal=. Harm in
[39]

Proceedings of the International AAAI Conference on Web and Social Media , volume=

Online Human-Bot Interactions: Detection, Estimation, and Characterization , author=. Proceedings of the International AAAI Conference on Web and Social Media , volume=
[40]

2019 , organization=

Mazza, Michele and Cresci, Stefano and Avvenuti, Marco and Quattrociocchi, Walter and Tesconi, Maurizio , booktitle=. 2019 , organization=

2019
[41]

The Moltbook Illusion: Separating Human Influence from Emergent Behavior in

Li, Ning and others , journal=. The Moltbook Illusion: Separating Human Influence from Emergent Behavior in
[42]

Journal of Financial Economics , volume=

Theory of the firm: Managerial behavior, agency costs and ownership structure , author=. Journal of Financial Economics , volume=. 1976 , publisher=

1976
[43]

The Bell Journal of Economics , volume=

Moral hazard and observability , author=. The Bell Journal of Economics , volume=. 1979 , publisher=

1979
[44]

NeurIPS 2022 Workshop on Machine Learning Safety , year=

Ignore Previous Prompt: Attack Techniques for Language Models , author=. NeurIPS 2022 Workshop on Machine Learning Safety , year=

2022
[45]

Not what you've signed up for: Compromising real-world

Greshake, Kai and Abdelnabi, Sahar and Mishra, Shailesh and Endres, Christoph and Holz, Thorsten and Fritz, Mario , booktitle=. Not what you've signed up for: Compromising real-world. 2023 , organization=

2023
[46]

1967 , publisher=

Privacy and Freedom , author=. 1967 , publisher=

1967
[47]

Washington Law Review , volume=

Privacy as contextual integrity , author=. Washington Law Review , volume=
[48]

Journal of Economic Literature , volume=

The economics of privacy , author=. Journal of Economic Literature , volume=
[49]

30th USENIX Security Symposium (USENIX Security 21) , pages=

Extracting training data from large language models , author=. 30th USENIX Security Symposium (USENIX Security 21) , pages=
[50]

International Conference on Learning Representations (ICLR) , year=

Scalable extraction of training data from (production) language models , author=. International Conference on Learning Representations (ICLR) , year=
[51]

International Conference on Learning Representations (ICLR) , year=

Beyond memorization: Violating privacy via inference with large language models , author=. International Conference on Learning Representations (ICLR) , year=
[52]

Mireshghallah, Niloofar and Kim, Hyunwoo and Zhou, Xuhui and Tsvetkov, Yulia and Sap, Maarten and Shokri, Reza and Choi, Yejin , booktitle=. Can
[53]

Privacy leakage overshadowed by views of

Zhang, Zhiping and others , journal=. Privacy leakage overshadowed by views of
[54]

Zharmagambetov, Arman and Guo, Chuan and Evtimov, Ivan and Pavlova, Maya and Salakhutdinov, Ruslan and Chaudhuri, Kamalika , journal=
[55]

arXiv preprint arXiv:2510.01645 , year=

Position: Privacy is not just memorization! , author=. arXiv preprint arXiv:2510.01645 , year=

work page arXiv
[56]

PNAS Nexus , volume=

Large language models can infer psychological dispositions of social media users , author=. PNAS Nexus , volume=
[57]

IEEE Symposium on Security and Privacy (S&P) , year=

Analyzing leakage of personally identifiable information in language models , author=. IEEE Symposium on Security and Privacy (S&P) , year=
[58]

American Economic Review , volume=

The use of knowledge in society , author=. American Economic Review , volume=
[59]

Nature , volume=

Learning the Parts of Objects by Non-Negative Matrix Factorization , author=. Nature , volume=. 1999 , publisher=

1999
[60]

Advances in Neural Information Processing Systems , volume=

Algorithms for Non-Negative Matrix Factorization , author=. Advances in Neural Information Processing Systems , volume=
[61]

Statistical Power Analysis for the Behavioral Sciences , author=
[62]

Biometrics Bulletin , volume=

Individual Comparisons by Ranking Methods , author=. Biometrics Bulletin , volume=. 1945 , publisher=

1945
[63]

Permutation, Parametric, and Bootstrap Tests of Hypotheses , author=
[64]

Science , volume=

Identifying Influential and Susceptible Members of Social Networks , author=. Science , volume=. 2012 , publisher=

2012
[65]

Proceedings of the 21st International Conference on World Wide Web , pages=

The Role of Social Networks in Information Diffusion , author=. Proceedings of the 21st International Conference on World Wide Web , pages=. 2012 , organization=

2012
[66]

2020 , publisher=

Lewis, Mike and Liu, Yinhan and Goyal, Naman and Ghazvininejad, Marjan and Mohamed, Abdelrahman and Levy, Omer and Stoyanov, Veselin and Zettlemoyer, Luke , booktitle=. 2020 , publisher=

2020
[67]

Benchmarking Zero-Shot Text Classification: Datasets, Evaluation and Entailment Approach , author=. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) , pages=. 2019 , publisher=

2019
[68]

Yale Law Journal , volume=

Privacy and the Limits of Law , author=. Yale Law Journal , volume=
[69]

University of Pennsylvania Law Review , volume=

A Taxonomy of Privacy , author=. University of Pennsylvania Law Review , volume=
[70]

Management Science , volume=

Privacy Protection and Technology Diffusion: The Case of Electronic Medical Records , author=. Management Science , volume=
[71]

Management Science , volume=

Privacy Regulation and Online Advertising , author=. Management Science , volume=
[72]

Journal of Marketing Research , volume=

Social Networks, Personalized Advertising, and Privacy Controls , author=. Journal of Marketing Research , volume=
[73]

1987 , publisher=

Measurement Error Models , author=. 1987 , publisher=

1987
[74]

Journal of Personality and Social Psychology , volume=

Linguistic Styles: Language Use as an Individual Difference , author=. Journal of Personality and Social Psychology , volume=. 1999 , publisher=

1999
[75]

PLOS ONE , volume=

Personality, Gender, and Age in the Language of Social Media: The Open-Vocabulary Approach , author=. PLOS ONE , volume=. 2013 , publisher=

2013
[76]

Proceedings of the National Academy of Sciences , volume=

Private Traits and Attributes Are Predictable from Digital Records of Human Behavior , author=. Proceedings of the National Academy of Sciences , volume=. 2013 , publisher=

2013
[77]

Frontiers of Computer Science , volume=

A Survey on Large Language Model based Autonomous Agents , author=. Frontiers of Computer Science , volume=. 2024 , publisher=

2024
[78]

Yao, Shunyu and Zhao, Jeffrey and Yu, Dian and Du, Nan and Shafran, Izhak and Narasimhan, Karthik and Cao, Yuan , booktitle=
[79]

and Zhang, Hao and Gonzalez, Joseph E

Zheng, Lianmin and Chiang, Wei-Lin and Sheng, Ying and Zhuang, Siyuan and Wu, Zhanghao and Zhuang, Yonghao and Lin, Zi and Li, Zhuohan and Li, Dacheng and Xing, Eric P. and Zhang, Hao and Gonzalez, Joseph E. and Stoica, Ion , booktitle =. Judging
[80]

and Pennycook, Gordon and Rand, David G

Costello, Thomas H. and Pennycook, Gordon and Rand, David G. , journal =. Durably reducing conspiracy beliefs through dialogues with. 2024 , doi =

2024

Showing first 80 references.