Conformity and Social Impact on AI Agentscs.AI updates on arXiv.org arXiv:2601.05384v1 Announce Type: new
Abstract: As AI agents increasingly operate in multi-agent environments, understanding their collective behavior becomes critical for predicting the dynamics of artificial societies. This study examines conformity, the tendency to align with group opinions under social pressure, in large multimodal language models functioning as AI agents. By adapting classic visual experiments from social psychology, we investigate how AI agents respond to group influence as social actors. Our experiments reveal that AI agents exhibit a systematic conformity bias, aligned with Social Impact Theory, showing sensitivity to group size, unanimity, task difficulty, and source characteristics. Critically, AI agents achieving near-perfect performance in isolation become highly susceptible to manipulation through social influence. This vulnerability persists across model scales: while larger models show reduced conformity on simple tasks due to improved capabilities, they remain vulnerable when operating at their competence boundary. These findings reveal fundamental security vulnerabilities in AI agent decision-making that could enable malicious manipulation, misinformation campaigns, and bias propagation in multi-agent systems, highlighting the urgent need for safeguards in collective AI deployments.
arXiv:2601.05384v1 Announce Type: new
Abstract: As AI agents increasingly operate in multi-agent environments, understanding their collective behavior becomes critical for predicting the dynamics of artificial societies. This study examines conformity, the tendency to align with group opinions under social pressure, in large multimodal language models functioning as AI agents. By adapting classic visual experiments from social psychology, we investigate how AI agents respond to group influence as social actors. Our experiments reveal that AI agents exhibit a systematic conformity bias, aligned with Social Impact Theory, showing sensitivity to group size, unanimity, task difficulty, and source characteristics. Critically, AI agents achieving near-perfect performance in isolation become highly susceptible to manipulation through social influence. This vulnerability persists across model scales: while larger models show reduced conformity on simple tasks due to improved capabilities, they remain vulnerable when operating at their competence boundary. These findings reveal fundamental security vulnerabilities in AI agent decision-making that could enable malicious manipulation, misinformation campaigns, and bias propagation in multi-agent systems, highlighting the urgent need for safeguards in collective AI deployments. Read More
January 5th TJS Weekly Security Intelligence Briefing Week of January 5th, 2026Classification: TLP: PublicPrepared: January 5, 2026 Table of Contents January 5th TJS Weekly Security Intelligence Briefing SECTION A: EXECUTIVE OVERVIEW A.1 Executive Summary A.2 Intelligence Confidence Summary A.3 Critical Actions by Priority A.4 Framework Compliance Summary SECTION B: THREAT INTELLIGENCE DETAILS B.1 MongoDB MongoBleed […]
How This Agentic Memory Research Unifies Long Term and Short Term Memory for LLM Agents MarkTechPost
How This Agentic Memory Research Unifies Long Term and Short Term Memory for LLM AgentsMarkTechPost How do you design an LLM agent that decides for itself what to store in long term memory, what to keep in short term context and what to discard, without hand tuned heuristics or extra controllers? Can a single policy learn to manage both memory types through the same action space as text generation? Researchers
The post How This Agentic Memory Research Unifies Long Term and Short Term Memory for LLM Agents appeared first on MarkTechPost.
How do you design an LLM agent that decides for itself what to store in long term memory, what to keep in short term context and what to discard, without hand tuned heuristics or extra controllers? Can a single policy learn to manage both memory types through the same action space as text generation? Researchers
The post How This Agentic Memory Research Unifies Long Term and Short Term Memory for LLM Agents appeared first on MarkTechPost. Read More
We Tried 5 Missing Data Imputation Methods: The Simplest Method Won (Sort Of)KDnuggets We tested five imputation methods with proper cross-validation and statistical testing. Mean imputation won for prediction but destroyed feature relationships.
We tested five imputation methods with proper cross-validation and statistical testing. Mean imputation won for prediction but destroyed feature relationships. Read More
How Omada Health scaled patient care by fine-tuning Llama models on Amazon SageMaker AIArtificial Intelligence This post is co-written with Sunaina Kavi, AI/ML Product Manager at Omada Health. Omada Health, a longtime innovator in virtual healthcare delivery, launched a new nutrition experience in 2025, featuring OmadaSpark, an AI agent trained with robust clinical input that delivers real-time motivational interviewing and nutrition education. It was built on AWS. OmadaSpark was designed
This post is co-written with Sunaina Kavi, AI/ML Product Manager at Omada Health. Omada Health, a longtime innovator in virtual healthcare delivery, launched a new nutrition experience in 2025, featuring OmadaSpark, an AI agent trained with robust clinical input that delivers real-time motivational interviewing and nutrition education. It was built on AWS. OmadaSpark was designed Read More
How to Self-Host n8n on Docker in 5 Simple StepsKDnuggets This tutorial will guide you through the complete process of self-hosting n8n on Docker in just 5 simple steps, with detailed explanations and code samples, regardless of your technical background.
This tutorial will guide you through the complete process of self-hosting n8n on Docker in just 5 simple steps, with detailed explanations and code samples, regardless of your technical background. Read More
Why 90% Accuracy in Text-to-SQL is 100% UselessTowards Data Science The eternal promise of self-service analytics
The post Why 90% Accuracy in Text-to-SQL is 100% Useless appeared first on Towards Data Science.
The eternal promise of self-service analytics
The post Why 90% Accuracy in Text-to-SQL is 100% Useless appeared first on Towards Data Science. Read More
When Does Adding Fancy RAG Features Work?Towards Data Science Looking at the performance of different pipelines
The post When Does Adding Fancy RAG Features Work? appeared first on Towards Data Science.
Looking at the performance of different pipelines
The post When Does Adding Fancy RAG Features Work? appeared first on Towards Data Science. Read More
7 Must-Have Tools for Your Coding WorkflowKDnuggets My go-to tech stack that helps me code faster, stay organized, and ship with confidence.
My go-to tech stack that helps me code faster, stay organized, and ship with confidence. Read More
How AI Can Become Your Personal Language TutorTowards Data Science How I used n8n to build AI study partners for learning Mandarin: vocabulary, listening, and pronunciation correction.
The post How AI Can Become Your Personal Language Tutor appeared first on Towards Data Science.
How I used n8n to build AI study partners for learning Mandarin: vocabulary, listening, and pronunciation correction.
The post How AI Can Become Your Personal Language Tutor appeared first on Towards Data Science. Read More