PreprintNot peer reviewedArtificial IntelligenceResearchers report alignment training raises confident but wrong LLM answersOriginal title: The Alignment Paradox: How Post-Training Amplifies Confident Hallucinations in Language ModelsA preprint argues that post-training alignment, not missing data, is the main cause of confidently wrong answers in language models.Qingjia Huang et al. · Added 22 Oct 2026
PreprintNot peer reviewedFrom BYDArtificial IntelligenceReserved chat-template tokens drive much of prompt-injection success, preprint findsOriginal title: Same Bytes, Different Authority: Reserved-Token Representations in Chat-Template Prompt InjectionA preprint reports that reserved control tokens, not just template text, account for most of chat-template prompt-injection success in LLM agents.Added 22 Oct 2026
PreprintNot peer reviewedArtificial IntelligencePreprint merges policy and value learning in one model for agent RL trainingOriginal title: SAPO: Single-Rollout Autoregressive Policy Optimization for Agentic Reinforcement LearningA preprint reports a single-model, single-rollout method that matches PPO's performance while cutting memory and runtime.Dayang Liang et al. · Added 22 Oct 2026
PreprintNot peer reviewedQuantum ComputingPreprint links quantum circuit wiring choices to whether training can actually find solutionsOriginal title: When Expressivity Is Not Enough: Discrete Routing Geometry in Variational Quantum CircuitsA preprint argues that discrete CNOT routing shapes not just what a quantum circuit can represent, but whether gradient descent can reach good solutions at all.Yu Wang · Added 22 Oct 2026
PreprintNot peer reviewedFrom Meta, LightOnArtificial IntelligenceResearchers report steering how much AI assistants trust user claimsOriginal title: TrustMI: Causally controlling how assistants trust their usersA preprint reports that adjusting internal activations can make language model assistants more or less willing to rely on unverifiable user claims.Added 22 Oct 2026
PreprintNot peer reviewedArtificial IntelligenceStudy finds LLMs struggle to spot deliberately disguised malicious smart contractsOriginal title: The Deception Delta: Adversarial Evaluation of LLM-Based Smart Contract Bytecode ForensicsA preprint reports that adversarial camouflage in smart contract bytecode sharply cuts large language models' ability to detect hidden drains.Timo Schefold · Added 22 Oct 2026
PreprintNot peer reviewedFrom OratomicQuantum ComputingResearchers report new qLDPC codes for high-rate, fast-decoded quantum processorsOriginal title: High-rate qLDPC processorsA preprint introduces mitten codes, qLDPC codes built from non-abelian groups that the authors say boost processor performance.Added 22 Oct 2026
PreprintNot peer reviewedQuantum ComputingResearchers test a renormalisation-based circuit ansatz on a 30-qubit quantum processorOriginal title: Emergent-Coupling-Based Ansatz Evaluated on a Superconducting Quantum ProcessorA preprint reports that a renormalisation-group-inspired ansatz outperforms standard hardware-efficient circuits on disordered quantum chains.Alina Joch et al. · Added 22 Oct 2026
PreprintNot peer reviewedArtificial IntelligenceResearchers propose Patcher, a defence against full-parameter malicious finetuning of open LLMsOriginal title: Defending Against Malicious Finetuning by Scaling Train-time Adversarial AttacksA preprint presents Patcher, an adversarial training method that reuses simulated attack vectors to defend LLMs against malicious finetuning.Haoming Wen et al. · Added 22 Oct 2026
PreprintNot peer reviewedFrom MicrosoftArtificial IntelligenceResearchers report RL framework boosts LLM creativity in scientific proposal writingOriginal title: Reinforcing Agentic Creativity in Scientific Ideation with Night ScienceA preprint reports that reinforcement learning taught language models to deviate from predictable reasoning, producing more varied and original research proposals.Added 22 Oct 2026
PreprintNot peer reviewedFrom QC WareQuantum ComputingPreprint finds matching quantum training loss does not ensure useful sample generationOriginal title: "Train classical, deploy quantum" requires rethinking generalizationA preprint reports that classically-optimised quantum generative models can match loss values yet generate very different shares of valid new samples.Added 22 Oct 2026
PreprintNot peer reviewedFrom IBM ResearchQuantum ComputingResearchers set criteria for when splitting quantum circuits across chips pays offOriginal title: Structural Conditions for Distributed Quantum AdvantageA preprint proposes three structural tests for distributed quantum advantage and verifies bridging two toric-code patches on real hardware up to 142 spins.Added 22 Oct 2026
PreprintNot peer reviewedArtificial IntelligenceResearchers report runtime that shares cloud compute across edge-cloud LLM splitsOriginal title: DySCo: Dynamic Sharding for Collaborative Edge-Cloud LLM Inference with Depth-Synchronized BatchingA preprint describes DySCo, a runtime that lets edge devices share partial LLM inference with the cloud more efficiently via depth-synchronized batching.Jingpo Xu et al. · Added 22 Oct 2026
PreprintNot peer reviewedArtificial IntelligencePreprint describes an AI agent that tunes RAG pipelines by diagnosing failuresOriginal title: Agentic AutoRAG: RAG Pipeline Optimization through Reasoning-Driven AgentsResearchers report an LLM-agent optimizer that diagnoses retrieval versus generation failures to tune RAG pipelines more cheaply and accurately.Lasse B. Strand et al. · Added 22 Oct 2026
PreprintNot peer reviewedQuantum ComputingPreprint examines when global search beats local search in quantum circuit optimisationOriginal title: Optimization Geometry of QAOA and Variational Quantum AlgorithmsA preprint studies how circuit design choices reshape the optimisation landscape of VQE and QAOA, and when global search outperforms local search.Vojtěch Novák et al. · Added 22 Oct 2026
PreprintNot peer reviewedQuantum ComputingResearchers demonstrate beam-splitter coupling between Kerr-cat qubit and transmonOriginal title: Experimental signatures of a $\hat{Z}\hat{X}$ beam-splitter interaction between Kerr-cat and transmon qubitsA preprint reports an experimental beam-splitter interaction linking a Kerr-cat qubit and a transmon, a step towards noise-biased error-correction ancillas.Josiah Cochran et al. · Added 22 Oct 2026
PreprintNot peer reviewedFrom ZTEArtificial IntelligenceA preprint describes G2PTQ, a refreshed gradient-and-Hessian method for compressing LLMsOriginal title: G$^2$PTQ: Improving LLM Post-Training Quantization with Generalized Gradient CompensationResearchers propose a post-training quantization method that updates gradient and Hessian estimates per block to reduce LLM compression errors.Added 22 Oct 2026
PreprintNot peer reviewedFrom SK hynixArtificial IntelligenceResearchers release trace dataset and simulator to study complex agentic AI systemsOriginal title: Characterizing How Complex Agentic AI Systems Handle General Tasks: A Trace-Based Simulation StudyA preprint introduces a token-level trace dataset and simulator for studying how multi-agent AI systems behave on general tasks.Added 22 Oct 2026
PreprintNot peer reviewedArtificial IntelligenceA preprint reports agents self-onboarded onto tasks cut costs sharply, but with variable resultsOriginal title: AutoOnboarder: Autonomously Onboarding LLM AgentsA preprint describes AutoOnboarder, a system that retrains and reconfigures agents for a task, reporting large cost cuts but variable outcomes.Tanay Arora et al. · Added 19 Oct 2026
PreprintNot peer reviewedArtificial IntelligencePreprint reports self-improving agent framework that adapts its own upgrade workflowsOriginal title: RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-AgentA preprint describes RSIFlow, a framework where a meta-agent diagnoses an AI agent's weak points and evolves its own improvement process.Huan Zhang et al. · Added 19 Oct 2026
