Church groups press Microsoft for 1% of data center costs; company stays vague
A coalition in Indiana says Microsoft's one-time donations fall far short of the tax breaks and community needs tied to its new data center.
AI & ML
Frontier models, research results, agents, and developer tooling.
OpenAI ties ChatGPT Work and Codex analytics to business value
OpenAI outlines how usage analytics for ChatGPT Work and Codex can tie AI adoption to measurable business outcomes.
NVIDIA Vera Rubin NVL72 Tops MLPerf Inference v6.1 Debut
NVIDIA's first MLPerf Inference results for Vera Rubin NVL72 position hardware and software co-design as central to inference economics.
AIUC Raises Series A to Underwrite AI Agents and Make Them Sue-able
AIUC applies insurance underwriting to AI agents, arguing that advanced AI should be financially accountable for its actions.
Apple Ships Siri AI Beta in iOS 27 and macOS Golden Gate 27
Apple's latest OS releases bring an AI-powered Siri beta to both platforms, alongside security fixes and design refinements.
AWS Adds GPU-Aware Inference Gateway to SageMaker HyperPod
A new managed EKS add-on routes large language model inference requests based on GPU availability, improving utilization on existing HyperPod clusters.
Google Co-Designs AI Tools with Designers for NYFW
Google is tailoring its Flow AI tools to fashion designers' workflows ahead of New York Fashion Week.
Old Anonymized Medical Data Can Skew AI Diagnoses
A Germany–UK study finds that decades-old anonymized patient records may still carry identifiable traits, causing AI to misdiagnose current patients.
SK hynix Launches AI-Focused Venture Capital Arm
The memory maker's new corporate venture capital brand targets AI ecosystem partnerships and future technology investments.
Agentic AI Adds Context and Coordination to Enterprise Automation
One analysis argues that context and coordination are the missing ingredients that could finally turn automation pilots into real transformation.
Enterprise AI Shifts From Tokenmaxxing to ROI-Driven Value
As cheap-token experiments fade, enterprise AI must now prove its worth through measurable returns, drawing lessons from digital accessibility.
Alibaba's Qwen3.8-Omni-Flash Cuts Video Tokens by 45.7%
A new omni-modal model from Alibaba combines audio-video understanding with agentic tool use and a 1M-token context window.
Roundup: 11 Open-Source Agent Harnesses for Local LLMs
A new roundup identifies 11 open-source agent harnesses verified to work with local LLM runtimes such as Ollama, LM Studio, and llama.cpp.
Salesforce Agentforce Adds Stress-Testing for Enterprise AI Agents
Agentforce aims to move autonomous agents from demo-grade prototypes to production-ready systems with synthetic testing and live optimization.
Quiet Day in AI: Newsletter Notes Lack of Major News
A prominent AI newsletter reports a day with little to report, highlighting a rare lull in the sector's usual churn.
Using AI vs. Building Around It: What Makes a Company AI-Native
A new piece asks where tool use ends and structural dependence on AI begins—and what that means for companies.
Korean Prompts Slip Past AI Safety Guards, Scale AI Benchmark Shows
A new bilingual English–Korean benchmark from Scale AI and the Korea AI Safety Institute found that Korean-language prompts grounded in local context consistently exposed safety weaknesses that English testing missed.
NATO-backed startup puts small AI models on drones for autonomous attacks
Scaleout is moving decentralized AI learning to military bases and drones, letting small onboard models identify and attack targets without a central connection.