AI

The Consensus Trap: Rescuing Multi-Agent LLMs from Adversarial Majorities via Token-Level Collaboration

Systematic Capability Benchmarking of Frontier Large Language Models for Offensive Cyber Tasks

Signal or Noise in Multi-Agent LLM-based Stock Recommendations?

The Open-Weight Paradox: Why Restricting Access to AI Models May Undermine the Safety It Seeks to Protect

Polarization and Integration in Global AI Research

Understanding Secret Leakage Risks in Code LLMs: A Tokenization Perspective

WebCompass: Towards Multimodal Web Coding Evaluation for Code Language Models

The Landscape of Agentic Reinforcement Learning for LLMs: A Survey

Safe for Whom? Rethinking How We Evaluate the Safety of LLMs for Real Users

Uncovering Logit Suppression Vulnerabilities in LLM Safety Alignment

Compliance of AI Systems

ReGA: Model-Based Safeguard for LLMs via Representation-Guided Abstraction

Rethinking Post-Unlearning Behavior of Large Vision-Language Models

Love, Lies, and Language Models: Investigating AI's Role in Romance-Baiting Scams

StealthGraph: Exposing Domain-Specific Risks in LLMs through Knowledge-Graph-Guided Harmful Prompt Generation

DynaWeb: Model-Based Reinforcement Learning of Web Agents

Stop Tracking Me! Proactive Defense Against Attribute Inference Attack in LLMs

SOK: A Taxonomy of Attack Vectors and Defense Strategies for Agentic Supply Chain Runtime

Josh Sirota: AI models must update frequently for business effectiveness, local hardware enhances data privacy, and proprietary solutions address task inefficiencies | TWIST

Josh Sirota discusses three critical trends in enterprise AI: the necessity for frequent model updates to maintain business relevance, the privacy advantages of deploying AI on local hardware rather than cloud infrastructure, and the value of proprietary solutions in solving specific task inefficiencies. These insights highlight a shift toward decentralized, privacy-first AI deployments in enterprise environments.

Anthropic releases Claude Design with Opus 4.7 amid US government pressure

Anthropic has released Claude Design alongside its Opus 4.7 model, demonstrating competitive market confidence despite regulatory pressures from the US government. The move signals Anthropic's intent to advance AI capabilities and market positioning in an increasingly scrutinized regulatory environment.

How to Ground a Korean AI Agent in Real Demographics with Synthetic Personas

The article discusses methods for grounding Korean AI agents using synthetic personas to improve demographic accuracy and localization. This approach addresses the challenge of making AI systems culturally relevant and demographically representative for specific markets, particularly in non-English speaking regions.

Moonshot AI’s Kimi K2.6 launch challenges Anthropic’s AI dominance

Moonshot AI has launched Kimi K2.6, a new AI model that directly competes with established players like Anthropic's Claude. The release signals intensifying competition in the large language model market, with potential implications for market consolidation and technological differentiation among AI providers.

Scaling Codex to enterprises worldwide

OpenAI has launched the Codex Transformation Partners program in collaboration with major consulting firms including Accenture, PwC, and Infosys to help enterprises implement and scale Codex across their software development workflows. This initiative represents a strategic shift toward enterprise adoption of AI-assisted coding tools through established technology partners.

Microsoft's GitHub grounds Copilot account sign-ups amid capacity crunch

Microsoft has halted new sign-ups for GitHub Copilot due to overwhelming demand exceeding system capacity. The move reflects surging interest in AI-powered coding tools but raises questions about infrastructure scalability and service accessibility during a critical growth phase.

Lose Your Job to AI? New York Lawmaker Proposes 'AI Dividend' Stimmy

New York congressional candidate Alex Bores has proposed an 'AI dividend' policy that would automatically trigger stimulus payments to Americans whose jobs are displaced by artificial intelligence adoption. The proposal represents an emerging political response to widespread concerns about AI-driven unemployment and income disruption.