<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>News and Updates on AI Solutions Wiki</title><link>https://ai-solutions.wiki/news/</link><description>Recent content in News and Updates on AI Solutions Wiki</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 16 Jun 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://ai-solutions.wiki/news/index.xml" rel="self" type="application/rss+xml"/><item><title>AWS Previews a FinOps Agent for Cloud Cost</title><link>https://ai-solutions.wiki/news/aws-finops-agent-preview/</link><pubDate>Mon, 15 Jun 2026 00:00:00 +0000</pubDate><guid>https://ai-solutions.wiki/news/aws-finops-agent-preview/</guid><description>Cloud bills are easy to run up and hard to read. That gap is what FinOps exists to close, and in the AWS Weekly Roundup on June 15, 2026, AWS introduced an AWS FinOps Agent in preview: an agentic assistant aimed at helping teams understand and control their cloud spend. The feature is in preview, so specifics will firm up over time, but it is worth knowing what FinOps is and why an agent is a natural fit for it, especially once AI workloads are on the bill.</description></item><item><title>Gemma 4 Lands on Amazon Bedrock</title><link>https://ai-solutions.wiki/news/gemma-4-amazon-bedrock/</link><pubDate>Mon, 15 Jun 2026 00:00:00 +0000</pubDate><guid>https://ai-solutions.wiki/news/gemma-4-amazon-bedrock/</guid><description>If you build on AWS, the practical question with any new model is rarely &amp;ldquo;is it good,&amp;rdquo; but &amp;ldquo;can I call it from the platform I already use, and which size fits my workload.&amp;rdquo; As of the AWS Weekly Roundup on June 15, 2026, Google DeepMind&amp;rsquo;s open Gemma 4 family is available on /tools/amazon-bedrock/, so you can reach it through the same managed API, security, and billing as every other Bedrock model, with no servers to run.</description></item><item><title>Why the US Restricted Anthropic's Fable 5 and Mythos 5</title><link>https://ai-solutions.wiki/news/anthropic-fable-mythos-us-restriction/</link><pubDate>Sun, 14 Jun 2026 00:00:00 +0000</pubDate><guid>https://ai-solutions.wiki/news/anthropic-fable-mythos-us-restriction/</guid><description>For anyone who builds on frontier AI, this is the scenario you plan for and hope never comes: a model you depend on is switched off by government order, for everyone, with no notice. On Friday, June 12, 2026, it happened. A US national security directive ordered Anthropic to cut off foreign access to its two most capable models, Claude Fable 5 and Claude Mythos 5. The scope was not practical to enforce one user at a time, so Anthropic disabled both models for everyone, worldwide.</description></item><item><title>EU AI Act: What Takes Effect on 2 August 2026</title><link>https://ai-solutions.wiki/news/eu-ai-act-august-2026-obligations/</link><pubDate>Thu, 07 May 2026 00:00:00 +0000</pubDate><guid>https://ai-solutions.wiki/news/eu-ai-act-august-2026-obligations/</guid><description>The EU AI Act reaches a major milestone on 2 August 2026. As the law is currently written, most of the regulation becomes applicable on that date: the rules for high-risk AI systems listed in Annex III, the full governance and enforcement framework, and the penalty regime. It is also the day the European Commission and its AI Office gain the power to enforce the general-purpose AI (GPAI) obligations that have technically been in force since 2 August 2025.</description></item><item><title>MCP Turns One: Async Tasks, a Registry, and Apps Built With OpenAI</title><link>https://ai-solutions.wiki/news/mcp-turns-one/</link><pubDate>Tue, 25 Nov 2025 00:00:00 +0000</pubDate><guid>https://ai-solutions.wiki/news/mcp-turns-one/</guid><description>One year after Anthropic introduced it, the Model Context Protocol (MCP) has gone from a single company&amp;rsquo;s idea to the default way AI agents connect to tools and data. On November 25, 2025, the project shipped the 2025-11-25 specification revision, marking the protocol&amp;rsquo;s first birthday with three changes that matter to anyone wiring a model up to the real world: an experimental Tasks primitive for long-running work, a simpler and more standards-based authorization model, and a growing official Registry of servers.</description></item><item><title>LangChain 1.0: The First Stable Release Consolidates on LangGraph</title><link>https://ai-solutions.wiki/news/langchain-1-0-release/</link><pubDate>Wed, 22 Oct 2025 00:00:00 +0000</pubDate><guid>https://ai-solutions.wiki/news/langchain-1-0-release/</guid><description>LangChain reached its first stable release on October 22, 2025, and it is more of a reset than an incremental bump. Both the Python and JavaScript packages hit 1.0 together, the sprawling collection of chains and helpers was trimmed down, and the framework now stands on a single agent abstraction (create_agent) running on the LangGraph runtime. A follow up, LangChain 1.1, landed on December 2, 2025. If you build LLM applications, this changes the recommended way to assemble agents and the contract you get back from models.</description></item><item><title>Amazon Bedrock AgentCore Goes GA, Then Adds Policy, Evaluations, and Payments</title><link>https://ai-solutions.wiki/news/bedrock-agentcore-general-availability/</link><pubDate>Mon, 13 Oct 2025 00:00:00 +0000</pubDate><guid>https://ai-solutions.wiki/news/bedrock-agentcore-general-availability/</guid><description>Building an agent in a notebook is easy. Running it in production, securely, with memory that survives restarts, identity it does not control, and a record of what it did, is the hard part. Amazon Bedrock AgentCore is AWS&amp;rsquo;s answer to that gap: a set of managed services for deploying and operating agents built with any framework, model, or protocol. It reached general availability on October 13, 2025, and over the following eight months AWS layered on the governance and transaction pieces that production teams actually ask for.</description></item><item><title>Google's A2A Protocol Moves to the Linux Foundation</title><link>https://ai-solutions.wiki/news/a2a-protocol-linux-foundation/</link><pubDate>Mon, 23 Jun 2025 00:00:00 +0000</pubDate><guid>https://ai-solutions.wiki/news/a2a-protocol-linux-foundation/</guid><description>If you build with AI agents, the hard part is rarely a single agent. It is getting agents from different vendors and frameworks to talk to each other without bespoke glue for every pair. That is the gap the Agent2Agent (A2A) protocol aims to close, and on June 23, 2025, at Open Source Summit North America in Denver, Google donated A2A to the Linux Foundation for vendor-neutral governance. The new Agent2Agent project launched with Amazon Web Services, Cisco, Google, Microsoft, Salesforce, SAP, and ServiceNow as founding contributors.</description></item></channel></rss>