Every paper checked — nothing hidden

All papers

Every paper checked for brief #4 of 12–19 Oct 2026 — nothing is hidden. Brief papers first.

All17,313 In brief24 More (on-topic)3,333 Not relevant13,754 Not assessed87 Older352 Could not confirm93 Blogs with no paper266 All blogs284

Paper	Topic	Relevance
Verification with Transfer: Exact Information Frontiers and Their Price in Calls Establishes information bounds and query complexity frontiers for verifying answers when transferring hints from related source tasks. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Addresses abstract theoretical verification games, not practical AI safety, guardrails, or hallucination mitigation.
Verification with Transfer: Exact Information Frontiers and Their Price in Calls Establishes information bounds and query complexity frontiers for verifying answers when transferring hints from related source tasks. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Theoretical information-theoretic study of verification protocols, not LLM benchmarks or evaluation frameworks.
One Node, Two Roles: Simultaneous Contests for Validation and Attention in Rollups Proves modeling limits and incentive mechanisms for separating validation from attention in optimistic blockchain rollups. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on blockchain rollup validator incentives and cryptography, not AI reliability or safety.
Improved Local Leakage Resilience of Shamir Secret Sharing and Worst-Case Optimal Polynomial Intersection Analyzes local leakage resilience in Shamir secret sharing and bounds worst-case optimal polynomial intersection over finite fields. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Theoretical cryptography on Shamir secret sharing and polynomial intersection, not AI deployment safety.
HOMARD: Hammering Off-chip Memory via Aggregate power Rail Disclosure The paper demonstrates microarchitectural attacks that extract DRAM mapping and memory contents using low-resolution on-chip power rail measurements without timers. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Examines hardware microarchitectural DRAM power side-channel attacks, unrelated to AI safety or model reliability.
Enabling Extended SGX Enclaves inside Confidential Virtual Machines The paper designs a monitor architecture leveraging AMD SEV-SNP privilege levels to run secure process enclaves inside confidential virtual machines. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Covers hardware enclave virtualization (SGX/CVMs), not AI safety, hallucination, or LLM deployment guardrails.
PVBRF-ID: Private, Verifiable and Byzantine-Robust Federated Intrusion Detection for Oilfield Enterprise Networks The paper presents a two-server federated learning protocol using secret sharing and public verification to defend against poisoning attacks in industrial networks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies federated learning protocols for network security, not efficient or small language models.
PVBRF-ID: Private, Verifiable and Byzantine-Robust Federated Intrusion Detection for Oilfield Enterprise Networks The paper presents a two-server federated learning protocol using secret sharing and public verification to defend against poisoning attacks in industrial networks. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Evaluates a federated protocol on Edge-IIoTset, not proposing new general LLM evaluation benchmarks.
PVBRF-ID: Private, Verifiable and Byzantine-Robust Federated Intrusion Detection for Oilfield Enterprise Networks The paper presents a two-server federated learning protocol using secret sharing and public verification to defend against poisoning attacks in industrial networks. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Applies standard federated intrusion detection to enterprise networks without addressing LLM safety or deployment risks.
Byzantine-Resilient Federated Large Language Model via Hypergraph Message Passing The paper introduces a hypergraph autoencoder framework to filter malicious model updates and preserve Byzantine resilience in federated large language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses federated aggregation defense, not model compression, distillation, or efficient inference serving.
Byzantine-Resilient Federated Large Language Model via Hypergraph Message Passing The paper introduces a hypergraph autoencoder framework to filter malicious model updates and preserve Byzantine resilience in federated large language models. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Focuses on robust federated aggregation against Byzantine attacks, not benchmarking or evaluating LLMs.
Onions in the ATmosphere: Decentralizing Trust in Identity The paper describes ongoing approaches to decentralize identity trust by linking human-readable handles with self-authenticating cryptographic identifiers. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Addresses decentralized web identity and naming protocols, not AI safety or deployment.
Power, Performance, and Area of HPC-Based Malware Detectors: An FPGA and Open-PDK ASIC Study The paper benchmarks the power, area, and performance tradeoffs of implementing hardware-performance-counter malware detectors on FPGAs and ASICs. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on hardware-level FPGA/ASIC malware detectors, not AI system safety or reliability.
The False Allure of "Security by Obscurity": A Game-Theoretic Approach The paper builds a game-theoretic model to evaluate the trade-offs between system transparency and security by obscurity against strategic adversaries. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Presents game-theoretic modeling of general security by obscurity, not AI safety or deployment.
Vulnerable Models in the AI Supply Chain: An Empirical Study of Adoption, Propagation, and Usage Patterns The paper conducts an empirical study tracking how publicly flagged vulnerable AI models propagate across code repositories and downstream applications. Preprint	LLMs for code via arxiv · 12 Oct, 16:28	Not relevant Analyzes software supply chain adoption of vulnerable models, not LLMs writing or fixing code.
Sample-Preserving Search-to-Decision Reduction for Noisy Linear Equations over Large Moduli The paper introduces a sample-preserving search-to-decision reduction for noisy linear equations over large moduli with polynomial runtime. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Theoretical cryptography paper studying reductions for noisy linear equations, not AI safety.
AgentValve: Sybil-Resilient Payment Authorization for Permissionless AI Markets The paper presents a trust-network protocol using smart contracts to bound financial exposure and mitigate Sybil attacks in permissionless AI agent markets. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on economic Sybil attack resistance and payment routing for agent markets, not AI safety.
WaveMark: Robust Wavelet-Domain Image Watermarking for Diffusion Models The paper proposes embedding multi-scale wavelet-domain watermarks into initial diffusion noise to enable robust provenance tracking under image removal attacks. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on watermarking generative image outputs rather than multimodal reasoning or cross-modal models.
WaveMark: Robust Wavelet-Domain Image Watermarking for Diffusion Models The paper proposes embedding multi-scale wavelet-domain watermarks into initial diffusion noise to enable robust provenance tracking under image removal attacks. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Addresses diffusion model provenance watermarking, not industrial computer vision or defect inspection.
Post-Quantum Session Establishment for Matter over Thread: Message Size, Transport, and Protocol Options The paper quantifies message overhead and transport limitations when integrating post-quantum cryptography into the Matter over Thread IoT protocol. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on post-quantum network protocols for IoT devices, not AI safety or deployment.
Provably Subexponential Algorithms for Bounded-Error LWE with Few Samples The paper proves subexponential time bounds for solving bounded-error Learning with Errors given very few samples using an adapted Wagner-style algorithm. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Theoretical lattice cryptography paper on LWE, not AI safety or deployment reliability.
Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors The paper evaluates label-only backdoor defenses and demonstrates that inspecting representations in unpoisoned feature spaces improves poison detection accuracy. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Studies label-flipping backdoor attacks in classification, not industrial computer vision applications.
Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors The paper evaluates label-only backdoor defenses and demonstrates that inspecting representations in unpoisoned feature spaces improves poison detection accuracy. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on vision classifier auditing rather than multimodal vision-language architectures.
Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors The paper evaluates label-only backdoor defenses and demonstrates that inspecting representations in unpoisoned feature spaces improves poison detection accuracy. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Focuses on auditing backdoor attacks rather than general LLM evaluation benchmarks.
Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors The paper evaluates label-only backdoor defenses and demonstrates that inspecting representations in unpoisoned feature spaces improves poison detection accuracy. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies backdoor defenses, not small or efficient model compression or serving.
CAPE - Context-Aware Prioritization Engine for Vulnerability Scoring and Management The paper proposes a context-aware prioritization engine that incorporates system operational details to improve enterprise vulnerability scoring and remediation decisions. Preprint	LLMs for code via arxiv · 12 Oct, 16:28	Not relevant Proposes a cybersecurity vulnerability scoring engine rather than code synthesis or code LLM tooling.
CAPE - Context-Aware Prioritization Engine for Vulnerability Scoring and Management The paper proposes a context-aware prioritization engine that incorporates system operational details to improve enterprise vulnerability scoring and remediation decisions. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on enterprise software CVE prioritization rather than AI model safety, guardrails, or alignment.
The Limits of Query Monitoring against Distributed Model Extraction The paper evaluates the theoretical and empirical limits of using query monitoring to detect distributed model extraction campaigns among ordinary users. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Investigates defending against unauthorized model extraction, not techniques for model efficiency or compression.
The Limits of Query Monitoring against Distributed Model Extraction The paper evaluates the theoretical and empirical limits of using query monitoring to detect distributed model extraction campaigns among ordinary users. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Focuses on extraction detection monitoring rather than LLM benchmarking or evaluation frameworks.
Safety Graph: When Safety Responses Become Attack Surfaces in Cyber-Physical Systems The paper introduces a formal modeling framework combining attack graphs and fault trees to analyze cyber attacks triggered by industrial safety responses. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Investigates classical industrial control cyber-physical cascades using formal planning, not AI model reliability or safety.
Understanding Verification Vulnerabilities in Cross-chain Bridges: An Empirical Study The paper provides an empirical study of cross-chain bridge verification vulnerabilities and categorizes common flaws in cross-chain smart contract validation logic. Preprint	LLMs for code via arxiv · 12 Oct, 16:28	Not relevant Examines blockchain smart contract verification logic manually and via audits, not LLMs for code.
Understanding Verification Vulnerabilities in Cross-chain Bridges: An Empirical Study The paper provides an empirical study of cross-chain bridge verification vulnerabilities and categorizes common flaws in cross-chain smart contract validation logic. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Analyzes decentralized cross-chain bridge bugs, not AI model reliability, safety, or deployment.
Understanding and Mitigating Cross-Session Attacks via Persistent Artifacts in Agent Skill Composition The paper analyzes vulnerabilities from persistent artifacts shared across AI agent skill sessions and develops a runtime defense against resulting cross-session attacks. Preprint	LLMs for code via arxiv · 12 Oct, 16:28	Not relevant Analyzes multi-session agent skill interactions rather than code generation or software engineering tools.
LETHE: Defending against Unauthorized Image-to-Video Generation via Value Diversity Collapse The paper presents an adversarial perturbation defense that disrupts the value vector diversity in image-to-video diffusion models to prevent unauthorized generation. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on adversarial perturbations against image-to-video generative models rather than multimodal reasoning or understanding.
LETHE: Defending against Unauthorized Image-to-Video Generation via Value Diversity Collapse The paper presents an adversarial perturbation defense that disrupts the value vector diversity in image-to-video diffusion models to prevent unauthorized generation. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on defending generative video synthesis, not industrial computer vision or defect inspection.
ECGWmark: Deep Watermarking for Electrocardiogram Signals The paper proposes a deep convolutional, quantization-free watermarking framework to embed metadata into electrocardiogram signals without hurting diagnostic details. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Addresses 1D medical time-series signals rather than multimodal vision-language or cross-modal models.
ECGWmark: Deep Watermarking for Electrocardiogram Signals The paper proposes a deep convolutional, quantization-free watermarking framework to embed metadata into electrocardiogram signals without hurting diagnostic details. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on ECG biomedical signals, not computer vision for industrial inspection or manufacturing.
ECGWmark: Deep Watermarking for Electrocardiogram Signals The paper proposes a deep convolutional, quantization-free watermarking framework to embed metadata into electrocardiogram signals without hurting diagnostic details. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Deals with medical signal watermarking rather than AI deployment risks, jailbreaks, or hallucinations.
Secrets in Preamble: Practical and Robust PHY Information Hiding for Wi-Fi The paper introduces a method to embed hidden data inside Wi-Fi legacy short training fields without breaking packet detection. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on physical-layer Wi-Fi steganography and telecommunications, not AI safety or deployment reliability.
A Systematic Study of Security Enforcement at Android Inter-Component Boundaries The paper evaluates security enforcement across Android inter-component communication boundaries, identifying recurring enforcement gaps. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on Android operating system IPC security rather than AI safety or LLM reliability.
Decision-Focused Learning for Cyber-Physical Risk-Aware Power Grid Operations The paper introduces a decision-focused framework that aligns power grid cyber intrusion detection directly with operational cost minimization. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on classical power grid ICS intrusion detection rather than generative AI deployment safety.
AI Models Can Provably Hide Arbitrary Capabilities The paper shows that backdoor mechanisms can cryptographically hide arbitrary neural network capabilities until activated by secret triggers. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on hiding capabilities cryptographically rather than model efficiency or compression.
Structured Sentiment Analysis Using Sequence Labeling as Dependency Graph Parsing The paper models structured sentiment analysis as dependency graph parsing implemented through sequence labeling across five languages. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Addresses sentiment parsing without examining deployment safety risks or guardrails.
Structured Sentiment Analysis Using Sequence Labeling as Dependency Graph Parsing The paper models structured sentiment analysis as dependency graph parsing implemented through sequence labeling across five languages. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Uses simpler sequence labeling architectures, but does not study model quantization, distillation, or serving.
Structured Sentiment Analysis Using Sequence Labeling as Dependency Graph Parsing The paper models structured sentiment analysis as dependency graph parsing implemented through sequence labeling across five languages. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Evaluates existing sentiment benchmarks rather than contributing a new benchmark or evaluation method.
Structured Sentiment Analysis Using Sequence Labeling as Dependency Graph Parsing The paper models structured sentiment analysis as dependency graph parsing implemented through sequence labeling across five languages. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Studies text sentiment analysis without visual or multimodal modalities.
Structured Sentiment Analysis Using Sequence Labeling as Dependency Graph Parsing The paper models structured sentiment analysis as dependency graph parsing implemented through sequence labeling across five languages. Preprint	LLMs for code via arxiv · 12 Oct, 16:28	Not relevant Applies dependency graph parsing to sentiment analysis, not software engineering or code.
Structured Sentiment Analysis Using Sequence Labeling as Dependency Graph Parsing The paper models structured sentiment analysis as dependency graph parsing implemented through sequence labeling across five languages. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Presents sequence labeling for sentiment analysis rather than LLM agent systems.
Structured Sentiment Analysis Using Sequence Labeling as Dependency Graph Parsing The paper models structured sentiment analysis as dependency graph parsing implemented through sequence labeling across five languages. Preprint	Retrieval-augmented generation (RAG) via arxiv · 12 Oct, 16:28	Not relevant Does not address retrieval-augmented generation or external knowledge retrieval.
4-Tensor Attention Model for Semantic Physical Reality The paper proposes a four-tensor attention architecture that jointly normalizes attention over semantic and temporal fibers for scene prediction. Preprint	LLMs for code via arxiv · 12 Oct, 16:28	Not relevant Does not address programming, software engineering, or code generation.
4-Tensor Attention Model for Semantic Physical Reality The paper proposes a four-tensor attention architecture that jointly normalizes attention over semantic and temporal fibers for scene prediction. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Introduces an architectural model rather than an LLM evaluation benchmark.
4-Tensor Attention Model for Semantic Physical Reality The paper proposes a four-tensor attention architecture that jointly normalizes attention over semantic and temporal fibers for scene prediction. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on attention efficiency and state modeling, not safety or guardrails.
4-Tensor Attention Model for Semantic Physical Reality The paper proposes a four-tensor attention architecture that jointly normalizes attention over semantic and temporal fibers for scene prediction. Preprint	Retrieval-augmented generation (RAG) via arxiv · 12 Oct, 16:28	Not relevant Focuses on scene state transitions rather than retrieval-augmented generation systems.
4-Tensor Attention Model for Semantic Physical Reality The paper proposes a four-tensor attention architecture that jointly normalizes attention over semantic and temporal fibers for scene prediction. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Tests semantic states on text stories rather than multimodal vision-language models.
4-Tensor Attention Model for Semantic Physical Reality The paper proposes a four-tensor attention architecture that jointly normalizes attention over semantic and temporal fibers for scene prediction. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Proposes an attention mechanism for state prediction without multi-step LLM agent systems.
Thinking Inertia: LLMs Keep Thinking When Told Not To The paper defines evaluation metrics to show that language models continue generating reasoning traces even when instructed not to think. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Studies reasoning mode suppression in LLMs rather than multi-step autonomous tool use.
Thinking Inertia: LLMs Keep Thinking When Told Not To The paper defines evaluation metrics to show that language models continue generating reasoning traces even when instructed not to think. Preprint	Retrieval-augmented generation (RAG) via arxiv · 12 Oct, 16:28	Not relevant Does not involve retrieval mechanisms, search, or external grounding.
Thinking Inertia: LLMs Keep Thinking When Told Not To The paper defines evaluation metrics to show that language models continue generating reasoning traces even when instructed not to think. Preprint	LLMs for code via arxiv · 12 Oct, 16:28	Not relevant Does not investigate code generation, software development, or program repair.
Thinking Inertia: LLMs Keep Thinking When Told Not To The paper defines evaluation metrics to show that language models continue generating reasoning traces even when instructed not to think. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Analyzes text-only reasoning traces without multimodal capabilities.
DPPM: Dual-Path Parametric Memory for Personalized Language Models The paper introduces a dual-path parametric memory method that combines pooled history representations with sequential adapter updates. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Presents a memory mechanism rather than a new general LLM evaluation methodology.

12,121–12,180 of 13,754← Previous Next →