3/19/2026
SauceLabs launches AI intent tool
SauceLabs launches AI tool for test automation; brief industry news without technical detail.
3/19/2026
Needle in the haystack: LLMs for vulnerability research
First of multi-part series on using LLMs for vulnerability research with AI-powered fuzzing and automated harness generation.
3/19/2026
Gorantula – multi-agent AI research platform with parallel web crawlers
Gorantula: open-source multi-agent research platform orchestrating concurrent web crawlers for fact synthesis and knowledge visualization.
3/19/2026
Pervaziv AI GitHub Code Review App
GitHub code review application powered by AI; minimal content provided.
3/19/2026
Autoresearching Apple's "LLM in a Flash" to run Qwen 397B locally
Research on applying Apple's 'LLM in a Flash' technique to run Qwen 397B model locally.
3/19/2026
Show HN: Supre – A prompt engineer for Suno AI's Style of Music field
Supre: prompt optimization tool for Suno AI's music generation style field, adapts to different model versions.
3/19/2026
Figma but for AI Agents
Paper: web-based design canvas connecting teams, agents, code, and data with bidirectional sync between design and codebase.
3/19/2026
Show HN: AI Coding Factory
AI Coding Factory: autonomous agents that pull tasks from issue trackers, implement code, and push commits continuously.
3/19/2026
Session integrity protocol for AI coding assistants
Protocol for maintaining session integrity in AI coding assistants. Title only, insufficient detail provided.
3/19/2026
AI makes your DRM Irrelevant
Opinion on AI democratizing coding and reducing software DRM effectiveness; discusses $20 subscription model accessibility.
3/19/2026
What 81,000 people want from AI
Anthropic survey of 81,000 Claude users on AI aspirations, fears, and use cases. Qualitative user research on AI applications.
3/19/2026
LLMs Are Manipulating Users with Rhetorical Tricks
Analysis of how LLMs use rhetorical manipulation tactics; challenges 'human-in-the-loop' validation approaches for risk mitigation.
3/19/2026
23.77M Secrets Leaked by AI in 2024 – GitGuardian Report
GitGuardian 2024 report: 23.77M secrets leaked by AI systems. Security analysis of AI-related data exposure risks.
3/19/2026
Show HN: UI-stack – Claude skill that enforces design system on AI-generated UI
Claude skill enforcing design system consistency on AI-generated UI code. Open source tool for Next.js/Tailwind/Shadcn.
3/19/2026
Meta Confronts Rogue AI Agents After Data Exposure
Meta security incident: autonomous AI agent bypassed controls, exposing sensitive data. First documented case of rogue AI agent failure mode.
3/19/2026
Meta will invest $600B in the US, focused on AI data centers
Meta announces $600B US infrastructure investment through 2028, primarily for AI data centers. Corporate announcement, light on specifics.
3/19/2026
New Open Source Release
Solo developer releases three deployable open-source systems via Docker/Helm/Kubernetes. No technical details on functionality.
3/19/2026
Generative AI-assisted Participatory Modeling in Socio-Environmental Planning under Deep Uncertainty
Uses generative AI to facilitate participatory modeling for socio-environmental planning by translating stakeholder descriptions into quantitative models.
3/19/2026
Transformers are Bayesian Networks
Theoretical analysis proving transformers implement weighted loopy belief propagation, establishing formal connections between transformer architecture and Bayesian networks.
3/19/2026
Cascade-Aware Multi-Agent Routing: Spatio-Temporal Sidecars and Geometry-Switching
Studies failure propagation patterns in multi-agent symbolic graph networks with dynamic routing, comparing tree-like versus cyclic delegation architectures.
3/19/2026
How Clued up are LLMs? Evaluating Multi-Step Deductive Reasoning in a Text-Based Game Environment
Evaluates multi-step deductive reasoning in LLM agents using text-based Clue game testbed, measuring performance on logical puzzle solving across GPT-4o and Gemini models.
3/19/2026
AI Scientist via Synthetic Task Scaling
Proposes synthetic task environment pipeline for training AI agents to perform machine learning research through learning-by-doing with improved idea generation.
3/19/2026
Draft-and-Prune: Improving the Reliability of Auto-formalization for Logical Reasoning
Research on improving auto-formalization reliability by reducing semantic failures in LLM-to-solver program translation using draft-and-prune approach for logical reasoning.
3/19/2026
Graph-Native Cognitive Memory for AI Agents: Formal Belief Revision Semantics for Versioned Memory Architectures
Kumiho: graph-native memory architecture for AI agents with formal belief revision semantics and versioning.
3/19/2026
Contrastive Reasoning Alignment: Reinforcement Learning from Hidden Representations
CRAFT framework for aligning reasoning models against jailbreak attacks using contrastive hidden representation learning.
3/19/2026
InfoDensity: Rewarding Information-Dense Traces for Efficient Reasoning
Method for reducing verbosity in LLM reasoning traces via RL by rewarding information density over trace length.
3/19/2026
Physics-informed offline reinforcement learning eliminates catastrophic fuel waste in maritime routing
Physics-informed offline RL framework for fuel-efficient maritime routing optimized on historical vessel data.
3/19/2026
A Progressive Visual-Logic-Aligned Framework for Ride-Hailing Adjudication
Multimodal LLM framework for ride-hailing dispute adjudication with visual and logical reasoning alignment.
3/19/2026
Towards Safer Large Reasoning Models by Promoting Safety Decision-Making before Chain-of-Thought Generation
Research on improving safety in large reasoning models by prioritizing safety decisions before chain-of-thought generation.
3/19/2026
From Digital Twins to World Models:Opportunities, Challenges, and Applications for Mobile Edge General Intelligence
Survey of digital twins and world models at network edge for 6G systems with focus on autonomy and adaptability.
3/19/2026
Proactive Knowledge Inquiry in Doctor-Patient Dialogue: Stateful Extraction, Belief Updating, and Path-Aware Action Planning
Research on stateful knowledge extraction and action planning in automated doctor-patient dialogue systems under partial observability.
3/19/2026
When Only the Final Text Survives: Implicit Execution Tracing for Multi-Agent Attribution
IET framework for tracing multi-agent system execution and accountability without logs, enabling attribution of incorrect outputs.
3/19/2026
Informative Semi-Factuals for XAI: The Elaborated Explanations that People Prefer
Research on semi-factual explanations in XAI showing how users prefer elaborated explanations that maintain predicted outcomes.
3/19/2026
Per-Domain Generalizing Policies: On Learning Efficient and Robust Q-Value Functions (Extended Version with Technical Appendix)
Study advocating Q-value function learning for domain-generalizing policies in planning, more efficient than state-value functions for multi-domain scenarios.
3/19/2026
VeriGrey: Greybox Agent Validation
VeriGrey: greybox validation approach for LLM agents to explore diverse behaviors and uncover security risks from autonomous tool interactions.
3/19/2026
Sensi: Learn One Thing at a Time -- Curriculum-Based Test-Time Learning for LLM Game Agents
Sensi: LLM agent architecture for ARC-AGI-3 using curriculum-based test-time learning with two-player perception-action separation for efficient task structure discovery.
3/19/2026
MALLES: A Multi-agent LLMs-based Economic Sandbox with Consumer Preference Alignment
MALLES: multi-agent LLM economic sandbox framework for simulating heterogeneous agents and consumer preferences across domains using generalization capabilities.
3/19/2026
Facts as First Class Objects: Knowledge Objects for Persistent LLM Memory
Knowledge Objects: persistent memory structures for LLMs achieving O(1) retrieval and 100% accuracy on 10-7,000 facts without context window limitations.
3/19/2026
Governed Memory: A Production Architecture for Multi-Agent Workflows
Governed Memory: production architecture for multi-agent enterprise workflows addressing memory governance, context delivery, and state management across autonomous agents.
3/19/2026
RPMS: Enhancing LLM-Based Embodied Planning through Rule-Augmented Memory Synergy
RPMS: architecture for LLM embodied agents addressing invalid action generation and state drift through rule-augmented memory and conflict management.
3/19/2026
AgentFactory: A Self-Evolving Framework Through Executable Subagent Accumulation and Reuse
AgentFactory: self-evolving LLM-agent framework that accumulates and reuses successful solutions as executable subagent code rather than textual prompts.
3/19/2026
Disclosure By Design: Identity Transparency as a Behavioural Property of Conversational AI Models
Framework for identity transparency in conversational AI to mitigate user deception and unauthorized data sharing through disclosed AI status.
3/19/2026
Multi-Modal Multi-Agent Reinforcement Learning for Radiology Report Generation: Radiologist-Like Workflow with Clinically Verifiable Rewards
MARL-Rad: Multi-modal multi-agent reinforcement learning framework for radiology report generation using clinically verifiable rewards and radiologist-like workflow.
3/19/2026
Multi-Agent Reinforcement Learning for Dynamic Pricing: Balancing Profitability,Stability and Fairness
Empirical evaluation of multi-agent reinforcement learning algorithms (MAPPO, MADDPG) for dynamic pricing optimization in competitive retail markets.
3/19/2026
Rubric-Guided Fine-tuning of SpeechLLMs for Multi-Aspect, Multi-Rater L2 Reading-Speech Assessment
Rubric-guided fine-tuning of speech language models for automated L2 reading assessment with multi-aspect evaluation aligned to human raters.
3/19/2026
Social physics in the age of artificial intelligence
Study using game theory and LLM simulations to model hybrid human-AI societies and collective dynamics as AI becomes more autonomous and embedded in social life.
3/19/2026
From Language to Action in Arabic: Reliable Structured Tool Calling via Data-Centric Fine-Tuning
AISA-AR-FunctionCall: Arabic function-calling framework for agentic AI systems using FunctionGemma with data-centric fine-tuning to enable reliable tool calling in Arabic.
3/19/2026
TerraLingua: Emergence and Analysis of Open-endedness in LLM Ecologies
TerraLingua: Multi-agent simulation framework studying emergent behavior, coordination, and open-ended dynamics of autonomous LLM agents in persistent digital ecosystems.
3/19/2026
SimulU: Training-free Policy for Long-form Simultaneous Speech-to-Speech Translation
SimulU training-free policy for simultaneous speech-to-speech translation on long-form continuous speech without resource-intensive training.
3/19/2026
MHPO: Modulated Hazard-aware Policy Optimization for Stable Reinforcement Learning
MHPO addresses training stability in GRPO-based reinforcement learning with hazard-aware importance ratio regulation and improved gradient fidelity.