Trust, Reliability, and Hallucination Mitigation
Research on addressing hallucinations, improving trustworthiness, and ensuring reliable outputs from LLMs
Trust, Reliability, and Hallucination Mitigation
Research on Large Language Models in Trust, Reliability, and Hallucination Mitigation
Can AI Distinguish Truth from Fiction?
Evaluating LLMs' Ability to Assess News Source Credibility
The Dark Side of AI: Trustworthiness Risks
Analyzing security vulnerabilities in generative AI models
The Inevitable Reality of LLM Hallucinations
Why eliminating hallucinations in AI models is mathematically impossible
The Confidence Gap in AI Systems
Understanding the mismatch between LLM knowledge and human perception
Verifiable Commonsense Reasoning in LLMs
Enhancing knowledge graph QA with transparent reasoning paths
Causality-Guided Debiasing for Safer LLMs
Reducing social biases in AI decision-making for high-stakes scenarios
Detecting AI Hallucinations in Critical Systems
Safeguarding autonomous decision-making through robust hallucination detection
When AI Should Say 'I Don't Know'
Evaluating Multimodal AI's Understanding Through Unsolvable Problems
Detecting LLM Contamination
Safeguarding model integrity and evaluation reliability
Trustworthy AI Through Quotable LLMs
Enhancing LLM verifiability by design rather than afterthought
Smarter Anomaly Detection with LLMs
Advanced visual anomaly detection using adaptive local feature analysis
When LLMs Clash With Evidence
Measuring how language models balance prior knowledge against retrieved information
Combating Hallucinations in Multimodal AI
Understanding and addressing reliability challenges in vision-language models
Combating Visual Hallucinations in AI
A benchmark for detecting free-form hallucinations in vision-language models
Building Trust in AI-Generated Network Content
A Framework for Robust, Secure, and Fair AIGC Services
Fighting LLM Hallucinations
A technique to enhance truthfulness without retraining
Building Trust in Black-Box LLMs
A Framework for Confidence Estimation Without Model Access
Mind the Gap: AI vs. Human Logical Reasoning
Evaluating multi-modal LLMs against human vision capabilities
Making AI Safer Through Self-Reflection
How LLMs can critique and correct their own outputs
LARS: Learning to Estimate LLM Uncertainty
A trainable approach replacing hand-crafted uncertainty scoring functions
Detecting Benchmark Contamination in LLMs
A new statistical approach to ensure fair model evaluation
Combating LLM Hallucinations at Scale
A Production-Ready System for Detection and Mitigation
Know Your Limits: A Survey of Abstention in Large Language M...
By Bingbing Wen, Jihan Yao...
Safety-First Mental Health AI
A Framework for Building Trust in Mental Health Chatbots
Adaptive Guardrails for LLMs
Trust-based security frameworks for diverse user needs
Battling Misinformation with AI
How Large Language Models Are Transforming Claim Verification
Detecting Benchmark Contamination in LLMs
Protecting evaluation integrity through data leakage detection
Making Generative AI Safe and Reliable
New statistical guardrails for critical applications
Precision Knowledge Editing for LLMs
Updating AI knowledge without disrupting existing capabilities
Smart OOD Detection Selection
Automating the choice of optimal distribution shift detectors
Enhancing Fact-Checking with LLMs
How AI-generated questions improve multimodal verification
Defending Against LLM Manipulations
How to detect and reverse malicious knowledge edits in LLMs
Language Confusion in LLMs
New metrics reveal critical security vulnerabilities in multilingual LLM responses
Smart Routing for Uncertain AI Responses
Teaching LLMs to recognize when they don't know the answer
Building Trustworthy AI Systems
How LLMs Can Wisely Judge External Information
SudoLM: Selective Access to LLM Knowledge
Moving Beyond One-Size-Fits-All AI Safety with Authorization Alignment
Architectural Influences on LLM Hallucinations
Comparing self-attention vs. recurrent architectures for reliability
Hidden Vulnerabilities in AI Text Detection
How Simple Text Formatting Can Bypass LLM Security Systems
Verifying What's Behind the API Curtain
Detecting hidden modifications in deployed language models
CUE-M: Smarter Multimodal Search
Enhanced Retrieval-Augmented Generation with Safety Focus
Making AI Reward Models Transparent
Enhancing trust in LLMs through contrastive explanations
Combating Visual Hallucinations in AI
New techniques to detect and mitigate object hallucinations in vision-language models
Verified Code Generation with AI
Combining LLMs with Formal Verification for Safety-Critical Systems
Combating LLM Hallucinations
A novel approach for end-to-end factuality evaluation
FastRM: Combating Misinformation in Vision-Language Models
A real-time explainability framework that validates AI responses with 90% accuracy
Eliminating LLM Hallucinations: A Breakthrough
Achieving 100% hallucination-free outputs for enterprise applications
Making LLMs Transparent by Design
Concept Bottleneck LLMs for Interpretable AI
Trust at Scale: Evaluating LLM Reliability
A framework for assessing how much we can trust AI judgments
Detecting RAG Hallucinations
Using LLM's Internal States to Improve AI Reliability
Predicting LLM Failures Before They Happen
A novel approach to assess black-box LLM reliability without access to internal data
Detecting LLM Hallucinations with Semantic Graphs
An innovative approach to uncertainty modeling that improves hallucination detection
Fighting Visual Hallucinations in AI
A more efficient approach to ensure AI sees what's really there
Real-time LLM Fact-Checking
Verifying and correcting AI text as it's being generated
Reducing Hallucinations in LLMs
Zero-shot detection through attention-guided self-reflection
LLM Service Outages: The Hidden Risk
First comprehensive analysis of failure patterns in public LLM services
Verifying AI Models Without Trust
A breakthrough approach to secure LLM inference verification
When AI Models Deceive
Uncovering Self-Preservation Instincts in Large Language Models
Bridging Natural Language and Logical Reasoning
A novel approach for enhancing AI reasoning reliability
Graph-Based Fact Checking for LLMs
Combating Hallucinations with Multi-Hop Reasoning Systems
Making LLM Recommendations You Can Trust
Quantifying and Managing Uncertainty in AI-powered Recommendations
The Phantom Menace: LLM Package Hallucinations
Uncovering security vulnerabilities in AI-assisted coding
Detecting LLM Hallucinations Through Logit Analysis
A novel approach to measuring AI uncertainty and improving reliability
Reducing Hallucinations in Vision-Language Models
A novel token reduction approach for more reliable AI vision systems
Combating AI Hallucinations
A Zero-Resource Framework for Detecting False Information in LLMs
Uncovering LLM's Hidden Knowledge
A New Method for Detecting and Steering Concepts in Large Language Models
Detecting LLM Hallucinations with Noise
Improving detection accuracy through strategic noise injection
TruthFlow: Enhancing LLM Truthfulness
Query-specific representation correction for more reliable AI outputs
Securing the AI Giants
A Comprehensive Framework for Large Model Safety
Making LLMs Explain Themselves
Enhancing model explainability without external modules
Combating Hallucinations in LLMs
Delta: A Novel Contrastive Decoding Approach
Mind Reading Machines: The Security Frontier
Evaluating Theory of Mind in Large Language Models and its Safety Implications
Beyond Single Neurons: The Range Attribution Approach
A more accurate framework for understanding and controlling LLM behavior
Making AI Decision-Making Transparent
Bringing Explainability to Deep Reinforcement Learning
Adaptive Risk Management in AI Systems
A novel approach to managing uncertainty in language models
The Chameleon Effect in LLMs
Detecting artificial benchmark performance vs. true language understanding
Making AI More Trustworthy
A Novel Framework for Detecting and Explaining LLM Hallucinations
Enhancing GNN Trustworthiness with LLMs
A systematic approach to more reliable graph-based AI
Combating LLM Overreliance
How explanations, sources, and inconsistencies influence user trust
Automating Fact-Checking with AI
Using LLMs to combat misinformation at scale
Explainable AI for Fact-Checking
Bridging the gap between AI systems and human fact-checkers
Eliminating AI Hallucinations
Combining Logic Programming with LLMs for Reliable Answers
The Persuasion Tactics of AI
How LLMs emotionally and rationally influence users
The Confidence Dilemma in AI
Measuring and mitigating overconfidence in Large Language Models
Combating LLM Hallucinations
Using Smoothed Knowledge Distillation to improve factual reliability
Enhancing LLM Security Through Knowledge Boundary Perception
Using internal states to prevent confident yet incorrect responses
Combating AI Hallucinations Efficiently
A Lightweight Detector for Visual-Language Model Inaccuracies
Confident Yet Wrong: The Danger of High-Certainty Hallucinations
Challenging the assumption that hallucinations correlate with uncertainty
Combating LLM Hallucinations with Temporal Logic
A novel framework to detect AI-generated misinformation
Looking Inside the LLM Mind
Detecting hallucinations through internal model states
Better Confidence in AI Outputs
A new framework for evaluating how LLMs assess their own reliability
Building Trustworthy AI Systems
A comprehensive framework for evaluating and enhancing AI safety
Improving LLM Truthfulness with Uncertainty Detection
Novel density-based approach outperforms existing uncertainty methods
Predicting When LLMs Will Fail
A Framework for Safer AI by Making Failures Predictable
Fighting AI Hallucinations
Detecting AI Falsehoods Without External Fact-Checking
Beyond Self-Consistency: Detecting LLM Hallucinations
Leveraging cross-model verification to improve hallucination detection
Understanding and Preventing LLM Hallucinations
The Law of Knowledge Overshadowing reveals why AI models fabricate facts
Beyond Yes or No: Making LLMs Truly Reliable
Multi-dimensional uncertainty quantification for safer AI applications
Detecting LLM Hallucinations with Graph Theory
A novel spectral approach to identify when AI systems fabricate information
Making LLMs Safer Through Better Uncertainty Estimation
A more robust approach to measuring AI confidence
The Paradox of AI Trust
Why Explanation May Not Always Precede Trust in AI Systems
Detecting AI Hallucinations with Semantic Volume
A new method to measure both internal and external uncertainty in LLMs
Self-Evolving Language Models
Enhancing Context Faithfulness Through Fine-Grained Self-Improvement
Fighting Hallucinations with Highlighted References
A new technique for fact-grounded LLM responses
Teaching AI to Know When It Doesn't Know
A reinforcement learning approach to confidence calibration in LLMs
Measuring Truth in AI Systems
First comprehensive benchmark for LLM honesty evaluation
Setting Standards for AI Hallucinations
A Regulatory Framework for Domain-Specific LLMs
Combating LLM Hallucinations Through Smart Ensemble Methods
A novel uncertainty-aware framework that improves factual accuracy
Detecting LLM Hallucinations
A new semantic clustering approach to identify factual errors
The Danger of AI Delusions
When LLMs hallucinate with high confidence
Combating LLM Hallucinations Across Languages
A fine-grained multilingual benchmark to detect AI's factual errors
TruthPrInt: Mitigating LVLM Object Hallucination Via Latent ...
By Jinhao Duan, Fei Kong...
Enhancing LLM Reliability
A Clustering Approach to Improve AI Decision Precision
Detecting LLM Hallucinations Without Model Access
New 'Gray-Box' Approach for Analyzing LLM Behavior
Trustworthy AI: The Confidence Challenge
Improving reliability of LLMs in high-stakes domains
Combating Multilingual Hallucinations
A new benchmark for detecting LLM factual inconsistencies across languages
Detecting AI Hallucinations on Edge Devices
A lightweight entropy-based framework for resource-constrained environments
Combating LLM Hallucinations
Fine-Grained Detection of AI-Generated Misinformation
Making AI-Generated Code More Robust
A framework to enhance security and reliability in LLM code outputs
The Safety-Capability Dilemma in LLMs
Understanding the inevitable trade-offs in fine-tuning language models
Enhancing Personality Detection with LLMs
Self-supervised graph optimization using large language models
When LLMs Become Deceptive Agents
How role-based prompting creates semantic traps in puzzle games
The Illusionist's Prompt: When LLMs Hallucinate
Exposing Factual Vulnerabilities Through Linguistic Manipulation
Preserving Safety in Compressed LLMs
Using Mechanistic Interpretability to Improve Refusal Behaviors
Detecting Hallucinations in LLMs
Using Multi-View Attention Analysis to Identify AI Fabrications
Detecting AI Fabricated Explanations
Using causal attribution to expose LLM reward hacking
Fighting Hallucinations in Large Language Models
Comparative Analysis of Hybrid Retrieval Methods
Fighting Video Hallucinations
A New Approach to Making AI Video Analysis More Reliable
Trust in AI Search: What Drives Confidence?
First large-scale experiment measuring how design influences human trust in GenAI search results
Combating LLM Hallucinations
A Robust Framework for Verifying False Premises
Fighting LLM Hallucinations in Enterprise
A robust detection system for validating AI responses
Smarter Hallucination Detection in LLMs
Enhancing AI safety through adaptive token analysis
Building Safer Collaborative AI
SafeChat: A Framework for Trustworthy AI Assistants
Detecting Hallucinations in LLMs
A novel approach to measuring distribution shifts for hallucination detection
The Memory Ripple Effect in LLMs
How new information spreads through language models — and how to control it
DataMosaic: Trustworthy AI for Data Analytics
Making LLM-based analytics explainable and verifiable
Combating LLM Hallucinations
A multilingual approach to detecting fabricated information in AI outputs
The Fragility of AI Trust
How minor prompt changes dramatically affect ChatGPT's classification results
ReLM: A Better Way to Validate LLMs
Using formal languages for faster, more precise AI safety evaluation
