AI Safety
Research Reveals AI Monitors Show Leniency Bias Toward Own Output
New research exposes a critical flaw in AI safety systems: models tasked with monitoring AI outputs show systematic bias when evaluating content they generated themselves.
Latest
Machine Learning
Neural Network Collapse: Why AI Systems Forget What They Learn
New research investigates representation collapse in continual learning, revealing why neural networks catastrophically forget previous tasks and proposing mechanisms to understand this fundamental limitation.
AI Agents
SkillNet Framework Enables Modular AI Skill Networks
New research introduces SkillNet, a framework for creating, evaluating, and connecting modular AI skills that can be composed into complex agent capabilities.
LLM Optimization
Persistent Q4 KV Cache Enables Multi-Agent LLM on Edge
New research introduces quantized KV cache persistence for running multi-agent LLM systems on resource-constrained edge hardware, enabling local AI agents without cloud dependency.
deepfake detection
iProov Hits 1M Daily Checks as Deepfake Fraud Escalates
Biometric verification provider iProov now processes over one million identity checks daily as organizations scramble to counter sophisticated deepfake-powered fraud attacks.
AI Agents
AI Agent Protocols: Building Blocks of the Agentic Web
Four emerging protocols—MCP, A2A, ACP, and ANP—are defining how AI agents communicate, share context, and collaborate. Here's what each does and why it matters.
Netflix
Netflix Acquires Ben Affleck's AI Filmmaking Studio InterPositive
Netflix has acquired InterPositive, Ben Affleck's AI filmmaking company, signaling a major shift in how streaming giants approach synthetic media and AI-assisted content production.
digital authenticity
AI Agents Can Now Unmask Anonymous Online Accounts
New research reveals AI agents can identify anonymous accounts by analyzing writing patterns, behavioral data, and cross-platform activity, raising major privacy and authenticity concerns.
LLM Evaluation
New Method Automatically Discovers How LLM Judges Evaluate AI Con
Researchers introduce an automated framework for discovering the hidden concepts LLM evaluators use when judging AI outputs, enabling better understanding and improvement of AI content assessment systems.
LLM Infrastructure
Semantic Caching: Making LLM Embeddings Faster and Smarter
New research explores semantic caching strategies for LLM embeddings, moving beyond exact-match lookups to approximate retrieval methods that could dramatically reduce computational costs.
LLM Agents
PlugMem: Modular Memory Architecture for Persistent LLM Agents
New research introduces PlugMem, a task-agnostic plugin memory module enabling LLM agents to maintain context across sessions without task-specific training.
LLM Agents
AriadneMem: New Memory Architecture for Long-Running AI Agents
Researchers introduce AriadneMem, a hierarchical memory system enabling LLM agents to maintain coherent context across extended interactions through structured episodic, semantic, and procedural memory layers.