TechniqueRLHF / Alignment8 recent entries1 Oct 2026Refusals That Bend: Measuring and Predicting Task Malleability in Embodied VLM PlannersarXiv:2609.38971v1 Announce Type: cross Abstract: Embodied vision-language models (VLMs) are increasingly deployed as high-level planners for robots because they generalize across diverse environments→2 Oct 2026Video-Index: A Curated Meta-Benchmark for Video UnderstandingarXiv:2610.00960v1 Announce Type: new Abstract: A video benchmark should reward the capability it claims to measure, yet models can exploit answer options, question text, or partial visual evidence. W
TechniqueRAG8 recent entries28 Jul 2026Reason Before You Retrieve: Agentic Planning for Multi-modal RAGarXiv:2607.22643v1 Announce Type: new Abstract: Multimodal retrieval-augmented generation (mRAG) aims to answer image-text queries with external knowledge, but most existing systems still retrieve dir→28 Jul 2026Agent-UCT: Upper Confidence Bounds Applied to Trees for Agentic Workflow Optimization with Cost-AwarenessarXiv:2607.24162v1 Announce Type: new Abstract: Optimizing agentic workflows, such as retrieval-augmented generation (RAG) pipelines, requires navigating a combinatorial space of discrete component ch→14 Aug 2026Reconcile Once, Write Anytime: A Trust-Tiered Librarian and a Multi-Agent Writer for Drift-Free, Point-in-Time ResearcharXiv:2608.12984v1 Announce Type: cross Abstract: Long-form research reports generated by large language models drift, contradict themselves, and lose provenance: the same metric appears with differen→19 Aug 2026Authorization Before Context: A Model-Neutral Audience Boundary Against Cross-Audience Memory Leakage in Agentic SystemsarXiv:2608.17148v1 Announce Type: cross Abstract: A personal language agent learns a fact from one audience and may later place it in the prompt it assembles for another. This memory-to-context step i→1 Sept 2026T-MAP: Red-Teaming LLM Agents with Trajectory-aware Evolutionary SearcharXiv:2603.22341v2 Announce Type: replace-cross Abstract: While prior red-teaming efforts have focused on eliciting harmful text outputs from large language models (LLMs), such approaches fail to capt→10 Sept 2026Building Trustworthy Graph-Agentic RAG for Social Good: Architectures, Failure Propagation, and Assurance by ConstructionarXiv:2609.06391v1 Announce Type: new Abstract: Graph-agentic retrieval-augmented generation combines structured evidence with adaptive controllers that can plan retrieval, traverse relations, verify →14 Sept 2026Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity?arXiv:2606.23189v2 Announce Type: replace-cross Abstract: Computer-use agents (CUAs) now act on a user's behalf across personal applications such as email, calendars, and to-do lists. This cross-appli→2 Oct 2026MOMAT: Mixture of Multiple Atlases for Low-Power Jailbreak Defense of Quantized LLMsarXiv:2610.01058v1 Announce Type: cross Abstract: Quantized large language models are increasingly deployed on edge devices for their low latency and energy efficiency. However, model quantization wea
TechniqueAgents8 recent entries2 Oct 2026Video-Index: A Curated Meta-Benchmark for Video UnderstandingarXiv:2610.00960v1 Announce Type: new Abstract: A video benchmark should reward the capability it claims to measure, yet models can exploit answer options, question text, or partial visual evidence. W→5 Oct 2026LPS-Bench: Benchmarking Safety Awareness of Computer-Use Agents in Long-Horizon Planning under Benign and Adversarial ScenariosarXiv:2602.03255v2 Announce Type: replace Abstract: Computer-use agents (CUAs) execute multi-stage tasks through tools, where an early unsafe decision can propagate to consequential actions. Evaluatin→7 Oct 2026Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal AttacksarXiv:2603.04364v2 Announce Type: replace-cross Abstract: Multimodal web agents that process both screenshots and accessibility trees are increasingly deployed to interact with web interfaces, yet the→8 Oct 2026Constrained-Action AI Remediation for SIEM/XDR via a NeMo-Guardrails ProxyarXiv:2610.09906v1 Announce Type: cross Abstract: Security Operations Centers (SOCs) for information technology and operational technology share one incident-response problem: a flood of correlated al→8 Oct 2026Adversarial Images Hijack Web Agents from Visual Grounding to Browser ExecutionarXiv:2610.09240v1 Announce Type: cross Abstract: Modern web agents built on large vision-language models process webpages, select relevant UI elements, and translate model outputs into browser action→10 Oct 2026Sources: Anthropic's AI agents submitted 20 visa applications via a form on the US State Department website; the applications were incomplete and not processed (New York Times)New York Times: Sources: Anthropic's AI agents submitted 20 visa applications via a form on the US State Department website; the applications were incomplete and not processed — The Philadelphia Polic→12 Oct 2026X-AgentRed: Red-Teaming Personal Agents in Evolving Multi-Service EcosystemsarXiv:2610.13886v1 Announce Type: new Abstract: Personal agents increasingly connect to services across the web and applications by using tools. As the effects of their actions accumulate and interact→12 Oct 2026LLMail-Inject: A Dataset from a Realistic Adaptive Prompt Injection ChallengearXiv:2506.09956v2 Announce Type: replace-cross Abstract: Indirect prompt injection attacks exploit a fundamental weakness of LLMs: executing unauthorized tasks from untrusted sources. Despite the pra
TechniqueFine-tuning8 recent entries4 Sept 2026RL-ADA: A World-Feedback Framework for Adversarially Robust Enterprise Dialogue AgentsarXiv:2609.02902v1 Announce Type: new Abstract: Deploying task-oriented dialogue agents in enterprise customer support faces a persistent annotation bottleneck: robust training requires labelled inter→15 Sept 2026Empirical Evaluation of Task-Based Permission Scoping Architecture for AI AgentsarXiv:2609.15422v1 Announce Type: new Abstract: AI agents are provisioned the same as employee-owned hosts in many enterprise settings with a static credential set fixed at deployment which includes a→21 Sept 2026SteganoBackdoor: Evading Data-Poisoning Defenses via Steganographic BackdoorsarXiv:2511.14301v4 Announce Type: replace-cross Abstract: Transformer-based models are highly susceptible to backdoor attacks via supervised fine-tuning (SFT). To red-team existing data-poisoning defe→28 Sept 2026Why your startup needs open models alongside frontier APIsEvery week, I talk with founders who are building at an unbelievable pace. Teams are moving from inception to product-market fit faster than ever, with foundation models wired deeply into their core p→30 Sept 2026CounterSteer: Suppressing Indirect Prompt Injection with Activation SteeringarXiv:2609.36570v1 Announce Type: cross Abstract: Indirect prompt injection makes an LLM agent treat untrusted retrieved text as instructions. We present CounterSteer, an inference-time defense that s→6 Oct 2026Reflections and Fragments: Securing LLMs Against Sequential Mosaic AttacksarXiv:2610.05346v1 Announce Type: new Abstract: Self-play red-teaming improves language-model safety by pitting attacker and defender roles against each other in a zero-sum game. However, real adversa→7 Oct 2026Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal AttacksarXiv:2603.04364v2 Announce Type: replace-cross Abstract: Multimodal web agents that process both screenshots and accessibility trees are increasingly deployed to interact with web interfaces, yet the→9 Oct 2026Cited but Not Consulted: A Counterfactual Audit of Legal Chain-of-Thought FaithfulnessarXiv:2610.12361v1 Announce Type: new Abstract: Large language models increasingly justify legal decisions by naming the statute or precedent behind a verdict, treated as evidence that the decision fo
TechniqueMultimodal8 recent entries11 Sept 2026An Experimental Evaluation of Multimodal Prompt Injection Attacks on Agentic AI FrameworksarXiv:2609.09404v1 Announce Type: cross Abstract: Agentic AI frameworks let a language model plan, keep memory, and call tools that reach real files, mail, and services. Most of these agents also read→28 Sept 2026Why your startup needs open models alongside frontier APIsEvery week, I talk with founders who are building at an unbelievable pace. Teams are moving from inception to product-market fit faster than ever, with foundation models wired deeply into their core p→1 Oct 2026Refusals That Bend: Measuring and Predicting Task Malleability in Embodied VLM PlannersarXiv:2609.38971v1 Announce Type: cross Abstract: Embodied vision-language models (VLMs) are increasingly deployed as high-level planners for robots because they generalize across diverse environments→2 Oct 2026PhysVista: Benchmarking Physical Intelligence in VLMs via a Perception-Reasoning-Assessment LooparXiv:2610.00559v1 Announce Type: new Abstract: Vision-Language Models (VLMs) have shown strong multimodal reasoning capabilities, yet whether they truly capture the physical consistency underlying re→7 Oct 2026Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal AttacksarXiv:2603.04364v2 Announce Type: replace-cross Abstract: Multimodal web agents that process both screenshots and accessibility trees are increasingly deployed to interact with web interfaces, yet the→7 Oct 2026ATLAS-AL: Adaptive Trust-Region for Latent Adversarial Searches via Active LearningarXiv:2610.07323v1 Announce Type: cross Abstract: Security evaluation of learning-based systems requires more than just testing the system against a fixed collection of attacks. It requires adaptive m→8 Oct 2026Adversarial Images Hijack Web Agents from Visual Grounding to Browser ExecutionarXiv:2610.09240v1 Announce Type: cross Abstract: Modern web agents built on large vision-language models process webpages, select relevant UI elements, and translate model outputs into browser action→9 Oct 2026False Claims, Credible Images: A Red-Teaming Benchmark for Commercial Image GeneratorsarXiv:2610.11112v1 Announce Type: cross Abstract: Image-generation models can now produce text-rich, natural-looking visual artifacts that are hard to distinguish from real-world evidence, such as new
TechniqueSafety8 recent entries2 Oct 2026MOMAT: Mixture of Multiple Atlases for Low-Power Jailbreak Defense of Quantized LLMsarXiv:2610.01058v1 Announce Type: cross Abstract: Quantized large language models are increasingly deployed on edge devices for their low latency and energy efficiency. However, model quantization wea→5 Oct 2026LPS-Bench: Benchmarking Safety Awareness of Computer-Use Agents in Long-Horizon Planning under Benign and Adversarial ScenariosarXiv:2602.03255v2 Announce Type: replace Abstract: Computer-use agents (CUAs) execute multi-stage tasks through tools, where an early unsafe decision can propagate to consequential actions. Evaluatin→6 Oct 2026Reflections and Fragments: Securing LLMs Against Sequential Mosaic AttacksarXiv:2610.05346v1 Announce Type: new Abstract: Self-play red-teaming improves language-model safety by pitting attacker and defender roles against each other in a zero-sum game. However, real adversa→7 Oct 2026Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal AttacksarXiv:2603.04364v2 Announce Type: replace-cross Abstract: Multimodal web agents that process both screenshots and accessibility trees are increasingly deployed to interact with web interfaces, yet the→8 Oct 2026Constrained-Action AI Remediation for SIEM/XDR via a NeMo-Guardrails ProxyarXiv:2610.09906v1 Announce Type: cross Abstract: Security Operations Centers (SOCs) for information technology and operational technology share one incident-response problem: a flood of correlated al→8 Oct 2026Adversarial Images Hijack Web Agents from Visual Grounding to Browser ExecutionarXiv:2610.09240v1 Announce Type: cross Abstract: Modern web agents built on large vision-language models process webpages, select relevant UI elements, and translate model outputs into browser action→9 Oct 2026ReSI: Recursive Safety Improvement toward Resistant and Resilient AIarXiv:2610.12233v1 Announce Type: cross Abstract: Recursive self-improvement, the participation of AI systems in improving their own capabilities, is beginning to move from theoretical prospect to pra→12 Oct 2026DistressBench: Evaluating Multi-Dimensional Crisis Support in Large Language ModelsarXiv:2610.12758v1 Announce Type: new Abstract: Language models have been adopted faster than evaluation practices that reliably assess their risks, and a growing share of use is affective rather than