LLM Reliability

LLM: Large Language Model

It asks not only whether a model gets an answer right, but whether it knows when it may be wrong, follows constraints, uses evidence faithfully, and behaves consistently across repeated or multi-turn use. Researchers test these properties with adversarial and perturbation benchmarks, repeated-run and cross-language evaluations, domain-specific tasks, and audits of LLM judges and tool-using agents. Common approaches include estimating uncertainty from logits, hidden representations, or response agreement; selective abstention and conformal calibration; retrieval and symbolic verification; and training methods that reward grounded answers, correction, or hesitation. The central tension is that benchmark accuracy and verbal confidence often fail to predict reliability on individual cases, long tasks, or shifted inputs, while stronger checks and abstention trade coverage, latency, and cost for safer decisions.

Latest papers

802 papers116 papers in the last four weeks, up 18% on the four weeks before. 0.7% of all new papers.

Surprise Alone Learns Lies: Evidence-Gated Continual Learning for LLMs from Streaming Claims
Oct 29, 2026Abdelkader Metales, Miriam GrabenheinrichContinual Learning for LLMs LLM Reliability
caaise GmbH
Beyond LLM Wrappers: Context Engineering and the Three-Engine Architecture for Reliable Financial AI
Oct 29, 2026Pradeep Kumar GS, Ravi TanniruContext Engineering LLM Context Management
Finteract AI
When the Same Prompt Gives Different Answers (and the Same Answer Isn't Always Right): A Taxonomy of Failure Mechanisms, Testing, and Mitigation Strategies for LLM-Assisted Evidence Synthesis Read-and-Evaluate Tasks
Oct 29, 2026Carole Lunny, Ian Goodman, Stefano Berrone +12LLM Reliability LLM Evaluation
Cochrane Hypertension Review Group, University of British Columbia.
Beyond Passing Tests: A Systematic Review and Controlled Re-evaluation of Techniques for Improving the Quality of LLM-Generated Code
Oct 28, 2026Prashank KadamLLM Code Generation Code Generation Evaluation
Cerby Inc., USA
"Close enough?" An investigation into multiplication errors in LLMs
Oct 28, 2026William Bolton, Bezawit Tilaye, Huw DayNumerical Reasoning in Language Models LLM Reliability
School of Engineering Maths & Technology University of Bristol · School of Computer Science University of Bristol
Evaluating Causal Reasoning in LLMs: A Case Study in the Challenges of Procedural Generation
Oct 28, 2026Justin Clarke, Isha Karn, Patrick Murray +2Causal Reasoning in Language Models Language Model Reasoning Evaluation
University of Massachusetts Amherst
Calculation and rejection diverge in language models tested at known mechanical limits
Oct 28, 2026Shaoliang Yang, Jun WangLLM Reliability Refusal Behavior in Language Models
Santa Clara University Santa Clara, CA 95053, USA
Measured or Real? When Model-Derived Measurements Can Be Corrected or Compared
Oct 27, 2026Yifeng Li, Ziang NiLLM Evaluation LLM Reliability
University of Trento · Delft University of Technology
Understanding Large Language Models as Numerical Optimizers
Oct 27, 2026Lingyun Guo, Ali BereyhiLarge Language Model-Guided Optimization Numerical Reasoning in Language Models
Edward S. Rogers Sr. Department of Electrical and Computer Engineering (ECE), University of Toronto, Toronto, ON, Canada · Schwartz Reisman Institute (SRI) for Technology and Society
Beyond QASecClaw: Agent Scaling, Verifier Feedback, and Evidence Reliability in Software Assurance
Oct 27, 2026S M Abdulla Al Mamun, Akif Islam, Mohd. Ruhul AmeenLLM Agent Verification LLM Reliability
Department of Computer Science, The University of Manchester, United Kingdom · Department of Computer Science and Engineering, University of Rajshahi, Bangladesh · Department of Computer Science and Engineering, Varendra University, Bangladesh
Confidence Is Not a Promise: A Reference Audit of Typed Decision Models
Oct 26, 2026Pedro Carvalho BromLLM Reliability LLM Auditing
Department of Mathematics, Federal Institute of Brasília, Brasília, Brazil
MARQ: A Modular, LLM-Agnostic Requirement Quality Assessor Infused with Systems Engineering Knowledge
Oct 26, 2026Jannatul Shefa, Alejandro Salado, Paul Wach +1Requirements Engineering Knowledge Augmentation for Language Models
Grado Department of Industrial and Systems Engineering, Virginia Tech, Blacksburg, Virginia, USA · Department of Systems and Industrial Engineering, University of Arizona, Tucson, Arizona, USA
Instruction Capacity: Behavior and Degradation of Language Models under Controlled Load
Oct 26, 2026Hennadii KutomanovLLM Evaluation Instruction Following
Independent researcher
Quantifying Claim Smoothing by LLM Research Agents
Oct 26, 2026Tiasa Singha Roy, Manya Wadhwa, Greg DurrettLLM Reliability LLM Agent Reliability
New York University
High Scores, Hidden Failures: Evaluating Codebook-Based LLM Content Analysis
Oct 26, 2026Elena Senger, Yuri Campbell, Rob van der Goot +1LLM-Assisted Annotation LLM Reliability
MaiNLP, Center for Information and Language Processing, LMU Munich & MCML, Germany · Fraunhofer Institute for Systems and Innovation Research ISI, Germany · Department of Computer Science, IT University of Copenhagen, Denmark
What Outcome RL Buys: Finishing, Reliability, Rarely Reach
Oct 26, 2026Shan YangReinforcement Learning RL for Language Model Reasoning
Individual Researcher.
Position Matters More Than Channel: Talking a Language Model Out of a Correct Answer
Oct 26, 2026Xinyi NiLLM Prompting Knowledge Conflicts in Language Models
Brandeis University
Models Verify Sufficiency, Not Validity: Usable Falsehoods Evade the Checks that Gaps Trigger
Oct 25, 2026Naveen MadirajuLLM Reliability Language Model Reasoning Evaluation
Google, San Jose, CA, USA
Higher Reasoning Effort Improves Reference Verifiability and Changes Reasoning Summaries in GPT-5.5
Oct 25, 2026Melika Mobini, Andres Algaba, Vincent GinisLLM Reliability Language Model Reasoning Evaluation
Vrije Universiteit Brussel, Brussels, Belgium · Harvard University (SEAS), Cambridge, MA, United States
Can LLMs Trust Appropriately? Evaluating Calibrated Reliance on Imperfect Partners
Oct 25, 2026Mengdi Chu, Jintao HuangTrust in AI LLM Reliability
The Ohio State University.
Why Does Structured Output Fail? An Empirical Study of Prompt-Enforced Pydantic Schemas Across Nine LLMs
Oct 24, 2026Amit VermaSchema-Constrained Generation LLM Reliability
Independent Researcher, Phoenix, AZ, USA.
Trust Mechanisms for Offline Generative AI at the Edge: A Connectivity-Regime Framework and an Empirical Study on a Raspberry Pi 5
Oct 24, 2026Georgios Feretzakis, Alexandros PapaspyridisLLM Reliability RAG Security
When Generated Information Returns: Error Persistence in LLM-Mediated Information Ecosystems
Oct 24, 2026Yi LiuMisinformation Diffusion LLM Reliability
University of Science and Technology of China
Adaptive Transfer of Representational Signatures for LLM Reliability Detection
Oct 24, 2026Yongtong Li, Yixuan Florence Wu, Jiawei Zhang +1Language Model Error Detection LLM Reliability
Northwestern University · University of Wisconsin–Madison
Scoped Semantic Invariance as a Foundation for LLM Reliability
Oct 23, 2026Rosario Uceda-Sosa, Maria Chang, Karthikeyan Natesan Ramamurthy +1Reasoning Consistency in Language Models LLM Reliability
IBM T. J. Watson Research Center, New York, USA
Scaling Qualitative Analysis with Large Language Models: Development and Validation of the Smart Coder Framework for Human-LLM Collaborative Coding
Oct 23, 2026Cheng Tang, Kun Wang, Yaxuan Yang +4LLM-Assisted Annotation LLM Prompting
Department of Educational Psychology, University of Georgia, Athens, GA, USA · Rethink Learning Labs, Woodcliff Lake, NJ, USA · Department of Early Childhood/Elementary Education, Penn State Abington, Abington, PA, USA
How Much Do You Trust A Dog? Quantifying Hint-Following in Reasoning Language Models
Oct 23, 2026Anxin Guo, Chirag Pabbaraju, Sara Ahmadian +2LLM Sycophancy LLM Reliability
Northwestern University, Evanston, IL, USA · Northwestern University, Evanston, IL, USA; NSF-Simons AI Institute for the Sky (SkAI), Chicago, IL, USA · Stanford University, Stanford, CA, USA +2
Free Lunch for LLMs Confidence: The RL Critic Already Knows
Oct 22, 2026Shujin Wu, Heng JiConfidence Estimation in Language Models LLM Reliability
University of Illinois Urbana-Champaign
Who Is Sycophantic to Whom? Confident LLMs, Human Deference, and the Collapse of Trust Calibration
Oct 22, 2026Deepthika S, Pawni Yadav, B. SankarTrust in AI LLM Reliability
Department of Electronics and Communication Engineering, Amrita School of Engineering, Amrita Vishwa Vidyapeetham, Coimbatore, 641112, India · Department of Information Technology, Netaji Subhas University of Technology New Delhi, 110078, India · Department of Mechanical Engineering, Indian Institute of Science (IISc), Bangalore, 560012, India
Major Revision, Minor Accountability: LLMs as Reviewers of Published Articles in Top-Tier Journals
Oct 22, 2026Bastien Le Guellec, Su Hwan Kim, Raphael Bentegeac +1Scholarly Peer Review Automated Peer Review
CHU Lille, Lille, France · Technical University of Munich (TUM), Munich, Germany

Oct 29, 2026J/K move · Enter open · S save

Surprise Alone Learns Lies: Evidence-Gated Continual Learning for LLMs from Streaming Claims

Abdelkader Metales, Miriam Grabenheinrich

caaise GmbH

Continual Learning for LLMs LLM Reliability Belief Updating in LLMs Data Provenance in Language Models

Open paper AskPDF arXiv

Continual learning methods for large language models increasingly decide what to learn by surprise, keeping the inputs the model finds least predictable. This rule is unsafe when the stream contains lies. A planted falsehood about something the model knows has low probability under the model's belief, so it is highly surprising by construction, and a surprise-driven learner absorbs nearly every such lie, at a rate similar to random selection with the same budget. We study the problem with LoRA-adapted students (Qwen2.5-1.5B, Qwen2.5-3B and Llama-3.2-3B) that learn from a stream of true and false claims about invented people and real CounterFact facts. We report four findings. (1) A verifier key, a judge that must approve each claim, removes lies to the extent that its verifier is accurate; real LLM judges are cautious and reject unsourced truths and even source-backed updates. (2) Calibrated regression tests on unrelated knowledge slow learning down without filtering out lies. (3) The original model's own prior flags claims that contradict what it knows, without a judge. It separates lies about known facts from true new facts (AUC 0.78-0.83 across three models) but not from legitimate updates of the same kind of fact (AUC 0.53). It saves 43% of judge calls, but an attacker with access to the model can choose lies that pass it. (4) If trusted sources exist that an attacker cannot plant, routing contradicted claims to evidence recovers legitimate updates and the combined system has the lowest poison rate among learned claims; when an attacker plants matching sources for 30% of the lies, the system ends up worse than the judge alone. The acceptance signals answer different questions, and none of them decides on its own whether a claim is true.