A long and bitter legal battle between tech billionaires Elon Musk and Sam Altman has culminated in victory for the OpenAI boss. Musk has vowed to appeal the verdict. The trial revealed insights into big tech and the global AI race.

A long and bitter legal battle between tech billionaires Elon Musk and Sam Altman has culminated in victory for the OpenAI boss. Musk has vowed to appeal the verdict. What did the trial reveal about big tech and the global AI race? Guardian US tech and power reporter Nick Robins-Early discusses.

PocketOS founder Jeremy Crane used Claude for routine DB maintenance, but the AI agent accidentally deleted the production database and all backups. Railway managed to recover the data. The incident highlights systemic weaknesses: overly broad tokens, non-expiring credentials, and lack of sandboxing. AI amplifies existing issues rather than being the root cause. Lessons include principle of least privilege, credential expiry, human-in-the-loop, and world models.

Sony debuts a special edition pair of 1000X-series headphones, striking a balance between modern style and function.

The risks of agentic AI are no longer theoretical. Agents connected to external tools are taking destructive, irreversible actions in production. Databricks is launching fine-grained access control, policy enforcement, and full audit trail for MCP tools via Unity Catalog, preventing such incidents and ensuring traceability.

Databricks launched Lakewatch at RSAC, an open agentic SIEM on Lakehouse, bringing AI threat detection where data lives. The article argues that CMOs must prioritize security infrastructure because AI-powered attacks outpace human response, and data breaches severely damage customer trust and brand health.

Databricks introduces the first context engineer certification, validating skills to design, manage, and govern AI context. The certification covers system prompts, retrieval, memory, tool integration, context window management, and governance. Beta exam available at Data + AI Summit.

This post explains how to use Amazon Nova Sonic, Amazon Bedrock AgentCore, and Strands BidiAgent to build scalable, maintainable voice agents. It details three architectural patterns: direct tool invocation via AgentCore Gateway, sub-agents for additional reasoning, and session segmentation for ultra-low latency. Each pattern's trade-offs and best practices are analyzed to help organizations deliver fast, natural, and reliable voice experiences.

This post demonstrates how to extend Kiro CLI's conversational memory by implementing a custom Model Context Protocol (MCP) server that integrates with Amazon Bedrock AgentCore Memory. Kiro CLI allows you to interact with AI agents directly from your terminal. Amazon Bedrock AgentCore Memory is a fully managed service that lets AI agents retain information from past interactions, enabling more intelligent and context-aware conversations. By implementing a custom MCP server, Kiro CLI gains tools to store and retrieve conversation context, monitor memory usage, and manage the underlying Bedrock Agent Core Memory infrastructure.

Amazon SageMaker Feature Store announces three new capabilities: native AWS Lake Formation integration, additional Apache Iceberg table properties, and SageMaker Python SDK v3 support, to simplify security governance and cost optimization for ML feature pipelines.

Programmatic tool calling (PTC) is a paradigm shift that reduces latency and token usage by having models generate Python code to orchestrate tool calls in a sandboxed environment. This post covers three implementation approaches on Amazon Bedrock: a self-hosted Docker sandbox on ECS, a managed solution using Bedrock AgentCore Code Interpreter, and an Anthropic SDK-compatible proxy path.

Valkey 9.1 is now generally available with a focus on improved efficiency, modularity, and ecosystem tooling for production workloads. It reduces per-key memory usage by up to 10% without tuning, integrates Valkey Search into the core for full-text and vector search, adds database-level ACLs and automated TLS reloading, and introduces CLUSTERSCAN for cluster-wide scanning. Valkey Admin reaches GA, and GLIDE client adds C# and PHP support. AI agents are used to automate backporting and bug detection.

Local AI search for Mac video & photo libraries.

This upcoming ebook, sponsored by Heroku by Salesforce, addresses the high failure rate of agentic AI projects and the critical role of a solid data foundation. It offers developers a guide to building agentic AI applications that tap into enterprise-wide knowledge using Salesforce Data 360, Agentforce, and Heroku. Key topics include why AI projects fail, the importance of unstructured data, retrieval-augmented generation (RAG), and streamlined integration.

DualCam AI enables recording from front and rear cameras simultaneously, ideal for video calls, vlogging, and more. Available on Product Hunt with a discussion thread.

Local-first AI memory bridge between browser chats and IDEs, enabling seamless context sharing without cloud dependency.

Databricks announces AI Spend Controls in Unity AI Gateway, providing proactive budget alerts to manage AI workload costs across users, workspaces, use cases, and accounts. Features per-user, per-use case, per-workspace, and per-account budgets, integrated with Unity Catalog and Databricks budgets.

Cowork is an autonomous agent inside Claude Desktop with direct folder access, enabling it to plan, execute, and deliver real work without step-by-step guidance.

Databricks announces LLM Guardrails in Unity AI Gateway (beta) to protect AI workloads from security and compliance risks through pre-built and custom guardrails, integrated with the lakehouse for observability.

Databricks announces expansion of Unity AI Gateway with LLM guardrails, cost controls, payload logging for MCP, and service policies to help teams govern AI agents at scale.

AVTR-1 is a new open weights model that generates uncanny AI avatars in real time, now open source.

Cloudflare has integrated with Anthropic's Claude Managed Agents to provide a fast, isolated execution environment for autonomous code delivery. This means builders can scale agent workflows globally while strictly controlling access to private backends and easily customizing their agent’s tools and runtimes.

MIT researchers developed EnergAIzer, a rapid prediction tool that estimates AI workload power consumption on specific processors in seconds with about 8% error, aiding data center energy efficiency.

The Perplexity alternative that organizes your research.

This article presents ten Python libraries that address key challenges in data engineering, organized into four areas: workflow orchestration, data ingestion, data quality, and storage/serialization. The libraries include Prefect, SQLMesh, dlt, Bytewax, PySpark, Great Expectations, Pandera, DuckDB, Polars, and Ibis, each with a brief description and learning resource.

Pulumi CEO Joe Duffy says that a year ago, VPs of infrastructure didn't think 'agentic AI' and 'infrastructure' belonged in the same sentence. Today, AI agents already drive 20% of all operations on Pulumi's platform, up from virtually zero a year ago, and Duffy thinks it could be 100% in the future. The company is shipping a bundle of capabilities for the 'agentic infrastructure era,' including free ephemeral accounts, a CLI makeover for agents, and extended Neo integration. Pulumi's use of general-purpose programming languages gives it an edge over Terraform's HCL for AI agents. Case studies include a healthcare customer clearing 400,000 compliance violations with Neo and migrating 500 Terraform workspaces in one hour.

Kimi WebBridge is a browser extension from Moonshot AI that enables local AI agents to directly control Chrome or Edge browsers, automating tasks such as page navigation, form filling, data extraction, and multi-step workflows while using existing login sessions and ensuring data stays local.

The company has AI handle most alerts, while retaining human oversight for nuanced situations.

Databox MCP connects your business data to AI assistants like Claude and ChatGPT, allowing you to ask questions in plain language and get answers grounded in real metrics and context.

Companies are racing to leverage AI for productivity, but most pilot projects fail. Investing in open standards like Agent Skills, MCP, and plugins protects against vendor lock-in and lowers switching costs. AI artifact catalogs help organizations turn individual wins into shareable institutional knowledge, enabling cross-team and agent reuse.