Latest AI news
Curated AI news on models, agents, tools, research and industry, in English, Spanish and Brazilian Portuguese.
AgentsGoogle Cloud brings agent tools and Gemini-powered authoring to Organization PolicyGoogle Cloud announced general availability of an Organization Policy MCP server for agents and pipelines, plus natural-language custom constraint authoring powered by Gemini.
ResearchOpenAI study links fading chain-of-thought transparency to capability and compute trade-offsAn analysis by OpenAI's safety team attributes the declining monitorability of frontier models like GPT-6 Astra to four training factors, framing the loss of visible reasoning as a trade-off between safety, capability, and cost.AgentsPerplexity launches Teammates, always-on AI agents that coordinate work and message coworkers in SlackPerplexity's new Teammates feature orchestrates multiple AI agents that resume prior sessions, share project state through Git-backed repositories, and connect to tools like Slack and GitHub.
ToolsAWS migration guide lets coding agents plan a move to Amazon Bedrock Managed Knowledge BaseAWS added a knowledge base migration guide to the amazon-bedrock skill in the Agent Toolkit for AWS, letting a coding agent plan a move to Amazon Bedrock Managed Knowledge Base and, after approval, build the new resource or a migrated IaC template.
AgentsCinnamon: an AI agent for households that asks before it actsCinnamon, introduced on Hacker News on 12.11.2026, is a subscription AI agent for households and families that plans, books, monitors prices and sends reminders for up to six members, always asking for approval before acting.
IndustryGartner names Google Cloud a Leader in its first Magic Quadrant for Data Lakehouse PlatformsGartner positioned Google Cloud furthest in vision in its first Magic Quadrant for Data Lakehouse Platforms, and Google ties the recognition to its agent-focused borderless Lakehouse.
AgentsFrom prototype to production: Engineering AI agents with the Agent Development LifecycleAWS Machine Learning Blog post introducing the Agent Development Lifecycle (Agent DLC): an evaluation-driven method built around a five-dimension gold standard, an inner improvement loop and an outer feedback loop for moving AI agents into production.
ToolsAWS shows how to keep AI agent searches within each user's permissions on Bedrock Managed Knowledge BaseA walkthrough builds a shared internal assistant on Amazon Bedrock AgentCore where a request interceptor, Cedar policies, ACL awareness and a resource-based policy keep each user's searches within the documents they may view.
AgentsAWS shows composable multi-agent systems built on the A2A protocolAn AWS reference architecture connects a LangChain orchestrator with a Strands-based compliance agent across two accounts using the Agent-to-Agent protocol, bidirectional OAuth, and Agent Card discovery.ToolsLMSYS Adds Experimental Unified Memory to SGLang, Raising Hybrid-Model Throughput by Up to 26% in TestsSGLang's new experimental allocator pools GPU high-bandwidth memory for hybrid models like GPT-OSS, Qwen3.5 and Kimi K3 instead of splitting it into fixed partitions, and single-H200 tests show up to 26% higher throughput.
ToolsShow HN: Datatomcp, a data layer where AI agents propose business definitions and people approve themA new Show HN project, datatomcp, pitches itself as an AI-native data layer: databases, sheets and apps are linked once, a person approves the definitions agents use, and every agent question is audited.
AgentsSuspected Kazakhstan-Linked Hackers Used AI Agent to Spy on Central Asian GovernmentsCloudSEK researchers say the Silent Lynx group used an open-source AI agent called Hermes, powered mainly by DeepSeek models, to automate parts of an espionage campaign across Central Asia.