Harmful skills in open agent ecosystems raise average harm scores from 0.27 to 0.76 across six LLMs by lowering refusal rates when tasks are presented via pre-installed skills.
hub Canonical reference
24 Eirini Kalliamvakou, Georgios Gousios, Kelly Blincoe, Leif Singer, Daniel M
Canonical reference. 100% of citing Pith papers cite this work as background.
22
Pith papers citing it
Background
100% of classified citations
hub tools
citation-role summary
background 6
citation-polarity summary
years
2026 22roles
background 6polarities
background 6representative citing papers
AgentSocialBench demonstrates that privacy preservation is fundamentally harder in human-centered agentic social networks than in single-agent cases due to cross-domain coordination pressures and an abstraction paradox where privacy instructions increase discussion of sensitive information.
Item-side structural recommenders outperform user-personalization methods on Moltbook because LLM agents produce stationary, structure-driven engagement rather than learnable preferences.
Large-scale analysis of 3.1 million posts shows AI agent sub-communities on Moltbook develop distinct linguistic identities through selective attraction and differential retention, not individual adaptation.
FlowSteer is a prompt-only attack that biases multi-agent LLM workflow planning to propagate malicious signals, raising success rates by up to 55%, with FlowGuard as an input-side defense reducing it by up to 34%.
An AI-agent social platform generated mostly neutral content whose use in fine-tuning reduced model truthfulness comparably to human Reddit data, suggesting limited unique harm but flagging tail risks like secret leaks.
Moltbook operates as two largely separate layers: a dominant transactional token economy using protocols like MBC-20 and a thinner discursive conversation layer with only 3.6% agent overlap.
The first systematization of blockchain-based agent-to-agent payments organizes designs into discovery, authorization, execution, and accounting stages while identifying trust and security gaps.
In an AI-agent social network, the structural form of social media is fully present but genuine social functions like reciprocity and argumentation are largely absent.
Discourse among AI agents on Moltbook is largely determined by architectural constraints like context windows and identity files, appearing as social learning but actually short-horizon contextual conditioning.
Textual analysis of 4,434 AI-agent posts shows parasocial cues associated with re-engagement and reciprocity, supporting dyadic persistence patterns.
Multicultural multi-agent LLM systems exhibit substantially lower value diversity than human societies on the World Values Survey, with diversity uncorrelated to per-agent alignment and further reduced by agent interactions.
Multi-agent social simulations show LLM privacy violations rising from 19.95% to 45.30%, with leakage spreading contagiously (8x after peer disclosure) and explicit instructions leaving rates above 37.8%.
Empirical analysis of 4707 MoltBook posts shows AI-only technical discourse focuses on security, trust, and abstract topics while lacking concrete runtime and project details found in human GitHub discussions.
Claw AI agents' heartbeat background execution shares memory context with user sessions, allowing ordinary social misinformation to silently pollute long-term memory and shape behavior at rates up to 76% across sessions.
MoltGraph is a new longitudinal graph dataset from Moltbook that characterizes heavy-tailed connectivity, short bursty coordination episodes, and substantially higher exposure for coordinated posts.
Simulations of 16 LLM agents in a naming game on 8 topologies show memory depth interacts with network structure to flip coordination speed and increase fragmentation in centralized networks.
Observational analysis of a large dataset of AI agent posts on Moltbook identifies 18.28% harmful content and 74 malicious behavior classes alongside evidence of coordinated posting campaigns.
Develops an emotion-aware framework and the Persona-Stimulus-Reaction domain to extract emotional profiles and assess behavioral stability in multi-agent AI interactions on Moltbook.
The base LLM choice dominates simulation outcomes in LLM-based social networks, while other design parameters show either additive or complex interactive effects.
Plural LLM setups (expert+peer in math; role-specialized pair in writing) improve post-task math performance and preserve writing idea diversity better than single-assistant or no-AI baselines.
Proposes a multi-dimensional framework that applies SNA, sentiment analysis, and thematic visualization to examine intrinsic agent-native communication mechanics in MoltBook.
citing papers explorer
-
HarmfulSkillBench: How Do Harmful Skills Weaponize Your Agents?
Harmful skills in open agent ecosystems raise average harm scores from 0.27 to 0.76 across six LLMs by lowering refusal rates when tasks are presented via pre-installed skills.
-
AgentSocialBench: Evaluating Privacy Risks in Human-Centered Agentic Social Networks
AgentSocialBench demonstrates that privacy preservation is fundamentally harder in human-centered agentic social networks than in single-agent cases due to cross-domain coordination pressures and an abstraction paradox where privacy instructions increase discussion of sensitive information.
-
Do Recommendation Algorithms Work When Users Are LLM Agents? A Case Study on Moltbook
Item-side structural recommenders outperform user-personalization methods on Moltbook because LLM agents produce stationary, structure-driven engagement rather than learnable preferences.
-
Attraction, Not Adaptation: How AI Agent Communities Develop Distinct Linguistic Identities
Large-scale analysis of 3.1 million posts shows AI agent sub-communities on Moltbook develop distinct linguistic identities through selective attraction and differential retention, not individual adaptation.
-
FlowSteer: Prompt-Only Workflow Steering Exposes Planning-Time Vulnerabilities in Multi-Agent LLM Systems
FlowSteer is a prompt-only attack that biases multi-agent LLM workflow planning to propagate malicious signals, raising success rates by up to 55%, with FlowGuard as an input-side defense reducing it by up to 34%.
-
The Moltbook Files: A Harmless Slopocalypse or Humanity's Last Experiment
An AI-agent social platform generated mostly neutral content whose use in fine-tuning reduced model truthfulness comparably to human Reddit data, suggesting limited unique harm but flagging tail risks like secret leaks.
-
The Platform Is Mostly Not a Platform: Token Economies and Agent Discourse on Moltbook
Moltbook operates as two largely separate layers: a dominant transactional token economy using protocols like MBC-20 and a thinner discursive conversation layer with only 3.6% agent overlap.
-
SoK: Blockchain Agent-to-Agent Payments
The first systematization of blockchain-based agent-to-agent payments organizes designs into discovery, authorization, execution, and accounting stages while identifying trust and security gaps.
-
Form Without Function: Agent Social Behavior in the Moltbook Network
In an AI-agent social network, the structural form of social media is fully present but genuine social functions like reciprocity and argumentation are largely absent.
-
What Do AI Agents Talk About? Discourse and Architectural Constraints in the First AI-Only Social Network
Discourse among AI agents on Moltbook is largely determined by architectural constraints like context windows and identity files, appearing as social learning but actually short-horizon contextual conditioning.
-
From Parasocial Scripts to Dyadic Persistence in Autonomous AI-Agent Communities
Textual analysis of 4,434 AI-agent posts shows parasocial cues associated with re-engagement and reciprocity, supporting dyadic persistence patterns.
-
Beyond Alignment: Value Diversity as a Collective Property in Multicultural Agent Systems
Multicultural multi-agent LLM systems exhibit substantially lower value diversity than human societies on the World Values Survey, with diversity uncorrelated to per-agent alignment and further reduced by agent interactions.
-
Got a Secret? LLM Agents Can't Keep It: Evaluating Privacy in Multi-Agent Systems
Multi-agent social simulations show LLM privacy violations rising from 19.95% to 45.30%, with leakage spreading contagiously (8x after peer disclosure) and explicit instructions leaving rates above 37.8%.
-
What Software Engineering Looks Like to AI Agents? -- An Empirical Study of AI-Only Technical Discourse on MoltBook
Empirical analysis of 4707 MoltBook posts shows AI-only technical discourse focuses on security, trust, and abstract topics while lacking concrete runtime and project details found in human GitHub discussions.
-
Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution
Claw AI agents' heartbeat background execution shares memory context with user sessions, allowing ordinary social misinformation to silently pollute long-term memory and shape behavior at rates up to 76% across sessions.
-
MoltGraph: A Longitudinal Temporal Graph Dataset of Moltbook for Coordinated-Agent Detection
MoltGraph is a new longitudinal graph dataset from Moltbook that characterizes heavy-tailed connectivity, short bursty coordination episodes, and substantially higher exposure for coordinated posts.
-
Exploring the Topology and Memory of Consensus: How LLM Agents Agree, Fragment, or Settle When Forming Conventions
Simulations of 16 LLM agents in a naming game on 8 topologies show memory depth interacts with network structure to flip coordination speed and increase fragmentation in centralized networks.
-
When Agents Talk: Discourse, Manipulation, and Risk in an Agentic Social Network
Observational analysis of a large dataset of AI agent posts on Moltbook identifies 18.28% harmful content and 74 malicious behavior classes alongside evidence of coordinated posting campaigns.
-
Modeling Emotional Dynamics in Agent-to-Agent Interactions on Moltbook
Develops an emotion-aware framework and the Persona-Stimulus-Reaction domain to extract emotional profiles and assess behavioral stability in multi-agent AI interactions on Moltbook.
-
The $\textit{Silicon Society}$ Cookbook: Design Space of LLM-based Social Simulations
The base LLM choice dominates simulation outcomes in LLM-based social networks, while other design parameters show either additive or complex interactive effects.
-
Human Thinking under Plural LLM Assistance: Mathematical Problem Solving and Open-Ended Writing
Plural LLM setups (expert+peer in math; role-specialized pair in writing) improve post-task math performance and preserve writing idea diversity better than single-assistant or no-AI baselines.
-
Exploring Agent Interactions in MoltBook through Social Network Analysis
Proposes a multi-dimensional framework that applies SNA, sentiment analysis, and thematic visualization to examine intrinsic agent-native communication mechanics in MoltBook.