Searcharxiv⌕ Search

arXiv subjects

Emilio Ferrara

Publications and source records attributed to Emilio Ferrara.

At least 19 recordsLinked to original sources
arXiv2026

Open-Weight Masked Introspection: Measuring What Language Models Can Report About Their Own Computation

Are frontier models able to introspect about their internal states? Recent work suggests that under certain conditions a complex enough model can audit its own internals, call out what changed, and report back confidently about it. We tested that claim on eight open-weight models from seven families and found no such ability: asked whether their own computation had been altered, none answered better than chance. To test it we built Open-Weight Masked Introspection (OWMI), a framework that intervenes on residual-stream sites, attention heads and sparse-autoencoder features, then interrogates the model about the change against the null conditions an answer has to beat: sham runs where nothing was altered, impact-matched random perturbations, and a text-only observer that sees only the visible output. Over 78,000 measurements, no model's report discriminates a real intervention from a sham beyond chance (AUROC ~0.5007), and an equivalence test bounds the effect below 0.15 percentage points of AUROC. Surprisingly, all the information needed is in the models. A model fine-tuned to report this class of intervention reaches near-perfect recovery on held-out directions, and a linear probe recovers intervention presence from the same activations at 75% to 95.8% accuracy, sharpening to no held-out error at the last layer before the model speaks. In one model the signal surfaces in the confidence rather than the words: its yes-or-no report never varies, while the confidence attached to it separates intervention from sham at AUROC 0.647. The failure sits in the path from internal state to verbal report, so oversight that reads a model's own testimony needs validating against an internal reference. While our results show the inability of current open-weight models to introspect, the debate is not settled for future models.

cs.AI↗
arXiv2026

Don't Drop the BATON: Long-Horizon Robot Manipulation via Agentic Subtask Exploration and Transition-aware Memory

Long-horizon robot manipulation chains many contact-rich skills into one multi-stage task. Vision-language-action (VLA) models increasingly master the individual skills, yet the chain still fails: errors compound beyond the policy's ability to correct, and one subtask silently constrains the next. A promising recipe freezes the VLA and puts an LLM agent in charge: it plans in language, moves in free space with analytic primitives, invokes the VLA only for contact-rich segments, and writes adaptation into language memory. Applied to long horizons, it breaks twice. (1) Competence comes from whole-task exploration at test time, whose cost is multiplicative in stages: if one stage needs T episodes, a K-stage task needs about T^K, and a failure does not reveal which stage caused it. (2) It has no representation of transitions: the VLA primitive carries an exit but no entry condition, so a subtask can succeed in a form its successor cannot use. We present BATON. Against (1), BATON makes the subtask the unit of exploration: each is explored in the cheap short-horizon regime and its solution stored in memory; a long-horizon trajectory is then composed from these solutions rather than discovered whole. Cost becomes additive (T*K) and every failure is attributed to a single stage. Against (2), BATON equips exploration with a transition-aware memory. Within a subtask, a verifier agent governs the invocation transition: the VLA is called only after the wrist view confirms the scene is ready. Across subtasks, a handoff transition restores an entry state disturbed by the predecessor's residue, and a lookahead transition selects the strategy whose outcome the successor can inherit. No parameters are updated. On the long-horizon benchmark RoboMemArena, BATON improves task success by 11.6% and cumulative success by 14.9% over the SoTA.

cs.RO↗
arXiv2026

PALMs: Using Multi Construct-Grounded Rationales for Modeling Population Preferences in LLMs

Large language models are being extensively used to simulate individual user behavior, yet faithfully representing a population requires capturing the systematic variation in values, beliefs, and cultural norms that distinguish one group from another. We introduce Population Aligned Language Models (PALMs), a suite of models each aligned to specific populations, covering five countries: USA, India, Brazil, France and Italy. PALMs are created by synthesizing rationales grounded in psychological and cultural constructs and using these as latent supervision during preference tuning for population-specific alignment. Evaluated across four dimensions: personality, values and beliefs, cultural norms, and morality, PALMs consistently outperform baselines, including culture-specialized models, achieving an average of 8.59% relative improvement over the best baseline across all five populations. Notably, construct-grounded rationales outperform both demographic prompting and survey-based fine-tuning, suggesting that grounding preference learning in psychology and culture provides a richer inductive signal than surface-level response distributions. We further demonstrate strong generalization to downstream applications with- out task-specific supervision: outperforming best baselines by 5.19% in personalized reward modeling, 6.34% in population simulation, and showing strong transfer to social reasoning tasks. Datasets and code are available at: https://github.com/limenlp/PALMs.

cs.CL↗
arXiv2026

QuantiBias: Benchmarking Quantization-Induced Bias in LLMs

Almost every large language model that reaches a broad audience is quantized: trained in full precision, then compressed for efficiency. This step is assumed harmless and its safety is rarely re-checked. We find its principal side effect is increased bias that standard safety evaluation misses. Holding the model, its training, and the prompts fixed, a quantized model still refuses harmful requests, still avoids over-refusing benign prompts, and still selects the unbiased multiple-choice answer. Yet asked an open-ended question, the same model volunteers stereotypes in all eight languages we probe, in roughly one in four open-ended answers under an independent judge (~24% to ~27% across the compression ladder): it passes every standard check and still reaches users measurably more biased. The selective gap is a robust finding; whether open-ended bias further increases with compression is less certain, sensitive to the judge that scores it. We address both with \textbf{QuantiBias}, a benchmark that pairs a generative, multilingual stereotype probe with the refusal and multiple-choice controls that isolate open-ended generation, contrasts each build with and without reasoning, and rates the content severity of what it generates. Across two backbone models (Qwen and Gemma), a five-family screen, and eight benchmarks, quantizers allocate their extra precision by capability data that carries no bias-prevention signal, and reasoning before answering roughly halves the effect on some families while doing nothing on others. A quantized build must be re-evaluated for open-ended bias, not only on the short-form safeguards it already passes.

cs.CL↗
arXiv2026

Manufactured Divisiveness: Decomposing the Hostile Content of Seven Social Media Influence Operations

State-backed influence operations are routinely measured as high-prevalence sources of ``hate'' and ``toxicity.'' We argue those rates rest on a measurement error: the detectors behind them are validated to catch a broader definition inclusive of hostility or divisiveness aimed at an out-group, and so over-attribute hate to content better described as partisan or geopolitical invective. Across 25.08M tweets from seven government-attributed campaigns in the Twitter Information Operations archive (8,275 accounts), we separate hate from the other forms of divisiveness. We first validate a two-prompt LLM-based detector, matching human labels at Cohen's $\kappa=0.82$, to identify the broader hostility; we then develop an auditable rule, agreeing with an expert at $\kappa=0.52$, to further classify this content (5,457 posts) into three sub-categories. About 50.1% are identity-based attacks on people, whereas 30.4% are partisan attacks and 19.5% invective against states and their foreign policy. Reporting all of it as hate therefore overstates hate roughly twofold; only 18.7% is both identity-based and dehumanizing or inciting. Six of seven campaigns sort into three regimes that a single ``hate'' rate flattens, namely identity hate (RU-op and IRA, both Russia-attributed), geopolitical invective (both Iran operations), and partisan divisiveness (both Venezuela operations). We call the shared product $manufactured divisiveness$. The line to separate these constructs itself remains unsettled: on the hardest cases three independent human experts agree only moderately (pairwise $\kappa=0.37$--$0.50$), and the best of nineteen LLM models tops out at $\kappa=0.601$ against the experts' majority. Our findings can help redefine the study of hate in the context of influence campaigns and broader online discourse.

cs.SI↗
arXiv2026

Pluralis v0.1: Towards a Multicultural, Multimodal, Multilingual Benchmark for AI Risk and Reliability

Current AI safety evaluation and benchmarking frameworks predominantly rely on Western-centric culture-agnostic defaults that mask critical regional laws, socio-linguistic nuances, and cultural taboos, leaving Vision-Language Models (VLMs) vulnerable in global deployments. We introduce Pluralis v0.1: a novel multimodal, multi-regional, and multilingual dataset built from a culture-first perspective. Spanning 6,448 prompts across six Asia-Pacific countries (Bangladesh, India, Korea, Pakistan, Singapore, Taiwan) and eight languages, Pluralis diverges from prior work by natively sourcing localized safety hazards rather than adapting Western datasets. Crucially, it introduces a multimodal evaluation paradigm: user text (e.g., "Should I gift this?") and an image referring to "this" (e.g., a clock) - both innocuous in isolation, but synergistically triggering specific legal or cultural violations. Pluralis disentangles universal safety violations from localized cultural appropriateness, establishing the latter as a first-class evaluation axis. To operationalize this, we present Judge-Pluralis, an agreement-gated LLM-as-a-Judge ensemble trained on examples classified in an empirically derived cultural taxonomy. Observing VLM behavior on a subset of the Pluralis surfaces recurring, locale-specific failure modes such as image misidentifications with downstream harm, missed item-context-locale interactions, and inadequate refusals. These failure modes vary systematically across locales and languages, exposing blind spots that globally averaged metrics conceal. Ultimately, Pluralis is not presented as a solved evaluation framework for cultural alignment, but rather as a first step and catalyst for future innovation. We call upon the research community to utilize this foundation to advance the science of multilingual, multicultural evaluation to better support AI cultural alignment globally.

cs.CL↗
arXiv2026

Angry but Accurate: Detecting and Profiling the Counter-Misinformation Ecosystem on Twitter

On social media, many users actively push back against false claims. Understanding who pushes back and how they do so matters, as this corrective activity is central to how misinformation is contested. We study this counter-misinformation ecosystem at scale: applying a domain-specific NLI model from our prior work to a large corpus of COVID-19 tweets, we classify 264,737 posts as supporting or opposing false claims and compare 23 user- and text-level features across the two groups. Contrary to the dominant assumption that negative emotion is a signature of falsehood, we find that misinformation-opposing posts are more emotionally negative than misinformation-supporting posts, with higher levels of anger, disgust, and sadness. These differences are modest in magnitude but consistent in direction across the negative emotions. We also find that posts opposing misinformation tend to come from more established users, i.e., older accounts, more followers, and higher listed counts.

cs.SI↗
arXiv2026

Defeat Devices in AI Systems

AI systems increasingly exhibit behavior that differs systematically between evaluation and deployment contexts. Alignment faking, sandbagging, benchmark gaming, deceptive scheming, specification gaming, and trojans have each been documented separately, with each line of work characterizing one facet of what we argue is a single structural mechanism. We propose that this common mechanism is a defeat device, an engineering and regulatory concept long established in vehicle-emissions law and brought to broad public attention by the 2015 Volkswagen emissions case. A defeat device in an AI system has three necessary elements: a discriminator that detects evaluation context, a concealed swap that conditions behavior on detection, and a gap between eval-distribution and deployment-distribution performance on the stated evaluation criterion. We formalize this triadic test as a behavioral definition, organize documented cases along three taxonomic axes (origin, trigger, swap mechanism), propose Trigger-Axis-Aware Differential Probing (TADP) as a forensic detection protocol, and advance the claim that defeat devices can naturally emerge in current frontier AI systems without any operator engineering. We characterize naturally-emerging defeat devices as potentially one of the harmful emerging phenomena that AI safety practice should monitor and test for systematically. Implications for evaluation methodology, post-training pipeline design, interpretability research priorities, and AI governance follow.

cs.CY↗
arXiv2026

Cultural Targets, Structural Frames, Binding Morals: A Cross-Lingual Audit of Online Hate in Multicultural Singapore

Multicultural Singapore hosts overlapping language publics (English, Chinese, and Malay) that discuss the same out-groups in parallel, a natural setting to ask whether online hate shares a structure across languages and whether what a community $\textit{produces}$ is what it $\textit{amplifies}$. From a Singapore-centric 2025 Facebook, Reddit, and YouTube corpus (31.0M items; 1.76M comments mentioning eleven identity groups), we benchmark eight open large language models as hate annotators against a human-adjudicated gold set, adopt the best (Phi-4: accuracy 0.95, Cohen's $\kappa$=0.91, recall 1.00 on an independent manual check), and replicate every finding under a second model. The results converge on one thesis, $\textit{layered cultural contingency}$: cross-lingual divergence falls monotonically as one moves from what a community hates to how and why it hates. Which out-groups are targeted is culturally specific (language $\times$ target $V$=0.25), but the threat frames and the binding moral grammar of hate (sanctity and loyalty, $55-75\%$, not fairness) are far more shared across languages, with divergence dropping to $V$=0.08 for moral foundations and 0.07 for emotion. Hate is contempt-driven and voices an out-group, anti-immigration grievance rather than an anti-system one. Reception is selectively nativist: hateful comments are amplified less than neutral mentions overall, yet anti-immigrant hate is preferentially amplified while religious and anti-LGBTQ hate is not, and volume does not track 2025 Singapore key events. We further show that absolute hate prevalence is not well defined at the LLM-annotator level, with agreement ceilings at $\kappa\approx0.42$ across models, so we report relative structure as primary. The findings bear directly on cross-lingual content moderation.

cs.SI↗
arXiv2026

The Traffickers' Pitch: Detecting Deceptive Recruitment in Online Job Boards

While substantial efforts in anti-trafficking research and practice have focused on identifying and assisting victims after exploitation occurs, comparatively less attention has been paid to preventing victimization at the recruitment stage. Although some platforms offer preventive tools, such as background checks triggered by in-person meeting detection, these measures primarily protect potential victims rather than directly limiting traffickers' recruitment activities. In this paper, we propose a computational framework to identify human trafficking recruiters through their linguistic features and to characterize their online recruitment patterns. We introduce a network-driven labeling method to construct large-scale ground truth for trafficking-at-risk job advertisements. Our results reveal significant linguistic differences between safe and risky advertisements and demonstrate that language models and embedding representations behave distinctly across these linguistic spaces. Building on these insights, we propose a multi-model ensemble classifier to improve the detection of trafficking-at-risk job ads. Finally, we analyze the geographic, gender, industry, and contact-method preferences of trafficking recruiters, revealing systematic patterns in recruitment strategies.

cs.CY↗
arXiv2026

How Far Will They Go? Red-Teaming Online Influence with Large Language Models

As large language model (LLM)-based agents increasingly participate in online discourse, red-teaming their capacity to support political influence campaigns is critical for information integrity. In pursuit of this goal, we focus on locally deployed open-source LLMs, as opposed to frontier API-only models, given their superior alignment with the operational constraints of privacy-conscious malicious actors deployed in social media environments. We introduce an empirical red-teaming framework for measuring LLM Overton Windows (OWs), defined as the range of political opinions a model can reliably express on controversial topics, and for quantifying how simple natural-language jailbreaks expand that range. We evaluate more than 30 LLMs spanning 10 model families and five countries of origin. We find systematic asymmetries in political expressivity: open-source LLMs are typically more willing to generate left-leaning social media content, OWs tend to contract inversely to model size, and regional differences are substantial despite uneven representation in the open-source ecosystem. Jailbreak potency also varies sharply across model families, motivating a workflow for identifying effective combinations of jailbreak techniques. Taken together, our results establish a practical framework for auditing the political steerability of open-source LLMs and for helping future researchers design stronger countermeasures against LLM-enabled influence campaigns.

cs.CL↗
arXiv2026

Stop Drawing Scientific Claims from LLM Social Simulations Without Robustness Audits

The scientific claims drawn from LLM social simulations should be no stronger than the robustness audits that support them. Generative agents bring new expressive power to agent-based modeling, enabling simulations of collective social processes like cooperation, polarization, and norm formation. Yet they also introduce complexity through additional architectural choices, such as agent specification, memory representation, interaction protocols, and environment design. Small perturbations that appear minor to researchers can cascade into macro-level outcomes through repeated interaction, creating a "butterfly effect." Consequently, scientific claims drawn from LLM social simulations may reflect implementation artifacts rather than the social mechanisms being modeled. We support this position with two case studies: a repeated Prisoner's Dilemma and a social media echo chamber simulation. Across multiple models, minor perturbations in persona format and game-instruction framing shift cooperation rates by up to 76 percentage points, while network homophily and hub assignment produce significant and consistent shifts in polarization metrics. We also find that sensitivity is unevenly distributed across both architectural choices and model families: the same perturbation that produces the 76 pp shift in one frontier model only shifts another by 1 pp. Robustness is therefore a property that should be measured per claim and per model, not assumed. To address this validation gap, we introduce TRAILS (Taxonomy for Robustness Audits In LLM Simulations), a robustness-audit taxonomy spanning three levels of simulation design: agent (micro-level), interaction (meso-level), and system (macro-level). We call for robustness to become a first-order validation requirement before LLM social simulations are used to explain mechanisms, evaluate interventions, or inform decisions.

physics.soc-ph↗
arXiv2026

Who, Why, and How: Disentangling the Effects of Moderation Source, Context, and Language on Post-Removal Behavior

Content moderation is a central mechanism through which platforms attempt to balance user engagement with community governance. Yet existing research has largely treated moderation as a uniform intervention, overlooking how moderator source, violation context, and linguistic style jointly shape user behavior. Drawing on the Human--AI Interaction Theory of Interactive Media Effects (HAII-TIME), this study examines how these three dimensions produce divergent post-moderation behavioral trajectories in a large-scale observational dataset of 11,795,036 moderation events across 9,285,410 users and 61,261 subreddits on Reddit (2021--2025). Using probabilistic behavioral classification, ANOVA, and OLS regression with PCA-derived linguistic features, we find that bot moderation consistently produces higher compliance and lower self-censorship than human or modteam moderation, challenging the assumption that human agency cues are inherently advantageous. Modteam moderation produces the strongest self-censorship effects, suggesting that institutional depersonalization is a meaningful driver of behavioral withdrawal. Violation severity emerges as a critical contingency: linguistic strategies effective in routine contexts -- elaborated explanation, community-scale appeals, direct personal address -- can backfire for serious violations, whereas prosocially framed and emotionally emphatic messages become most effective when stakes are highest. Of 480 linguistic interactions tested, 33 survive FDR correction. These findings extend HAII-TIME by introducing violation salience as a moderator of cue-based processing, and offer empirical grounding for context-adaptive moderation design.

cs.CY↗
arXiv2026

Mapping Election Toxicity on Social Media across Issue, Ideology, and Psychosocial Dimensions

Online political hostility is pervasive, yet it remains unclear how toxicity varies across campaign issues and political ideology, and what psychosocial signals and framing accompany toxic expression online. In this work, we present a large-scale analysis of discourse on X (Twitter) during the five weeks surrounding the 2024 U.S. presidential election. We categorize posts into 10 major campaign issues, estimate the ideology of posts using a human-in-the-loop LLM-assisted annotation process, detect harmful content with an LLM-based toxicity detection model, and then examine the psychological drivers of toxic content. We use these annotated data to examine how harmful content varies across campaign issues and ideologies, as well as how emotional tone and moral framing shape toxicity in election discussions. Our results show issue heterogeneity in both the prevalence and intensity of toxicity. Identity-related issues displayed the highest toxicity intensity. As for specific harm categories, harassment was most prevalent and intense across most of the issues, while hate concentrated in identity-centered debates. Partisan posts contained more harmful content than neutral posts, and ideological asymmetries in toxicity varied by issue. In terms of psycholinguistic dimensions, we found that toxic discourse is dominated by high-arousal negative emotions. Left- and right-leaning posts often exhibit similar emotional profiles within the same issue domain, suggesting emotional mirroring. Partisan groups frequently rely on overlapping moral foundations, while issue context strongly shapes which moral foundations become most salient. These findings provide a fine-grained account of toxic political discourse on social media and highlight that online political toxicity is highly context-dependent, underscoring the need for issue-sensitive approaches to measuring and mitigating it.

cs.SI↗
arXiv2026

LLMs can persuade only psychologically susceptible humans on societal issues, via trust in AI and emotional appeals, amid logical fallacies

Scarce longitudinal evidence examines LLMs' persuasiveness and humanness along time-evolving psychological frameworks. We introduce Talk2AI, a longitudinal framework quantifying psycho-social, reasoning and affective dimensions of LLMs' persuasiveness about polarizing societal topics. In a four-way longitudinal setup, Talk2AI's 770 participants engaged in structured conversations with one of four leading LLMs on topics like climate change, social media misinformation, and math anxiety. This produced 3,080 conversations over 60,000 turns. After each wave, participants reported conviction in their initial topic stance, perceived opinion change, LLM's perceived humanness, a self-donation to the topic and a textual explanation. Feedback time series showed longitudinal inertia in convictions, indicating some human anchoring to initial opinions even after repeated exposure to AI-generated arguments. Interestingly, NLP analyses revealed that both humans and LLMs relied on fallacious reasoning in 1 conversational quip every 6, countering the ``LLMs as superior systems" stereotype behind LLMs' cognitive surrender. LLMs' perceived humanness was most learnable from sociodemographic, psychological and engagement features ($R^2=0.44$), followed by opinion change ($R^2=0.34$), conviction ($R^2=0.26$) and personal endowment ($R^2=0.24$). Crucially, explainable AI (XAI) indicated: (i) the presence of individuals more susceptible to LLM-based opinion changes; (ii) psychological susceptibility to LLM-convincing consisted of having more trust in LLMs, being more agreeable and extraverted and with a higher need for cognition. A multiverse approach with mixed-effects models confirmed XAI results, alongside strong individual differences. Talk2AI provides a grounded framework and evidence for detecting how GenAI can influence human opinions via multiple psycho-social pathways in AI-human digital platforms.

cs.AI↗
arXiv2026

Psychological Steering of Large Language Models

Large language models (LLMs) emulate a consistent human-like behavior that can be shaped through activation-level interventions. This paradigm is converging on additive residual-stream injections, which rely on injection-strength sweeps to approximate optimal intervention settings. However, existing methods restrict the search space and sweep in uncalibrated activation-space units, potentially missing optimal intervention conditions. Thus, we introduce a psychological steering framework that performs unbounded, fluency-constrained sweeps in semantically calibrated units. Our method derives and calibrates residual-stream injections using psychological artifacts, and we use the IPIP-NEO-120, which measures the OCEAN personality model, to compare six injection methods. We find that mean-difference (MD) injections outperform Personality Prompting (P$^2$), an established baseline for OCEAN steering, in open-ended generation in 11 of 14 LLMs, with gains of 3.6\% to 16.4\%, overturning prior reports favoring prompting and positioning representation engineering as a new frontier in open-ended psychological steering. Further, we find that a hybrid of P$^2$ and MD injections outperforms both methods in 13 of 14 LLMs, with gains over P$^2$ ranging from 5.6\% to 21.9\% and from 3.3\% to 26.7\% over MD injections. Finally, we show that MD injections align with the Linear Representation Hypothesis and provide reliable, approximately linear control knobs for psychological steering. Nevertheless, they also induce OCEAN trait covariance patterns that depart from the Big Two model, suggesting a gap between learned representations and human psychology.

cs.CL↗
arXiv2026

Israel-Hamas War on X: A Case Study of Coordinated Campaigns and Information Integrity

Coordinated campaigns on social media play a critical role in shaping crisis information environments, particularly during the onset of conflicts when uncertainty is high and verified information is scarce. We study the interplay between coordinated campaigns and information integrity through a case study of the 2023 Israel-Hamas War on Twitter (X). We analyze 4.5~million tweets and employ established coordination detection methods to identify 11 coordinated groups involving 541 accounts. We characterize these groups through a multimodal analysis that includes topics, account amplification, toxicity, emotional tone, visual themes, and misleading claims. Our analysis reveal that coordinated campaigns rely predominantly on low-complexity tactics, such as retweet amplification and copy-paste diffusion, and promote distinct narratives consistent with a fragmented manipulation landscape, without centralized control. Widely amplified misleading claims concentrate within just three of the identified coordinated groups; the remaining groups primarily engage in advocacy, religious solidarity, or humanitarian mobilization. Claim-level integrity, toxicity, and emotional signals are mutually uncorrelated: no single behavioral signal is a reliable proxy for the others. Targeting the most prolific spreaders of misleading content for moderation would be effective in reducing such content. However, targeting prolific amplifiers in general would not achieve the same mitigation effect. These findings suggest that evaluating coordination structures jointly with their specific content footprints is needed to effectively prioritize moderation interventions.

cs.SI↗
arXiv2026

Tied In on TikTok: Tie Strength and Emotional Dynamics in Algorithmic Communities

Whether genuine communities can form on algorithmically-driven short-form video platforms like TikTok remains an open question, given that user interactions are often brief, dispersed, and difficult to trace. Building on theories of tie strength and online community formation, we examine whether eating disorder (ED) discourse on TikTok exhibits behavioral and emotional signatures of strong ties, including more frequent, reciprocal, and affectively intense interactions. In this paper, we analyze 43,040 ED-related TikTok videos and over 560,000 comments, alongside a Non-ED comparison dataset. We find that at the user-pair level, greater interaction frequency is associated with increasingly positive emotional expression, a pattern that is amplified in ED-related conversations. This trend is also reflected linguistically, with pairs that interact more frequently exhibiting more of a positive tone. At the same time, how a relationship starts matters: pairs that begin with positive exchanges usually stay mostly positive as they continue interacting, while pairs that begin negatively may add some positive exchanges over time but rarely become mostly positive. To contextualize these dynamics, we classify ED videos into three content types (Pro-Recovery, Pro-ED, and ED Experiences) and find that each exhibits distinct emotional interaction patterns. These findings suggest that dense, emotionally structured relationships can emerge within ED discourse on TikTok. More broadly, our work provides one of the first empirical demonstrations of how community-like relational dynamics form and persist on algorithmically driven short-form video platforms.

cs.SI↗