Today's trail

Australia’s own AI powers up to rival global giants

Visual Cipher Bench Tests LLM Agents on Long-Horizon Visual Reasoning Without Text Hints

Visual Cipher Bench evaluates multimodal LLM agents on 40 visual puzzles without text hints, exposing steep failure rates and rule discovery bottlenecks.

SailPoint: Rogue AI agents expose weaknesses in enterprise security

Endpoint protection platform comparison: Defender vs CrowdStrike Falcon vs SentinelOne on AI agent controls, prices from $59.99 and migration steps.

Understanding WordPress MCP, GripMCP, and AI Website Security

Discover how WordPress MCP and GripMCP empower AI assistants to securely manage your website, boosting efficiency and protecting your online presence.

NVIDIA CEO exit risks AI startup funding

NVIDIA CEO Jensen Huang abruptly resigned amid serial extramarital affair allegations, raising alarms across the global artificial intelligence (AI) startup sector. NVIDIA had previously led the expansion of the industry ecosystem by injecting massive funds into related startups.

Primer Framework Enables Instructor-Guided AI Simulations

Wharton researchers have released Primer, an open-source multi-agent AI framework enabling educators to create custom, voice-guided learning simulations.

Skill Adapter Personalizes and Routes Reusable LLM Agent Skills

On 12 November 2026, researchers debuted Skill Adapter, enabling LLM agents to adapt shared skills and route them dynamically during multi-step tasks.

Lara Mini Base Multilingual Pretraining Study

Anthropic allowed outside researchers from METR into its San Francisco headquarters for months, then let them publish a conclusion that its research…

Uber's Secure AI Framework Governs Autonomous AI Agents Across Their Lifecycle

Opinion: If the customer service is good, does it matter if it's AI?

Companies, not lawmakers, should decide how AI customer service is delivered, as long as satisfaction stays high and laws are followed.