News archive.
Earlier AI news and analysis, kept for reference. Current coverage lives in News & Analysis.
NVIDIA JetPack 7.2 Adds One-Command AI Agent Deployment for Edge Devices
JetPack 7.2 lets you deploy NemoClaw agents on Jetson hardware with a single bash command. New GPU partitioning, memory optimization skills, and a 20% performance boost on the 32GB Orin model are included.
JetBrains releases Mellum2, a 12B code-and-text model activating only 2.5B parameters per token
Mellum2 is a Mixture-of-Experts model optimized for routing, RAG, and sub-agent tasks in multi-model systems. The model achieves more than 2x faster inference than similarly-sized competitors while maintaining competitive benchmarks.
Anthropic files confidential S-1 with SEC ahead of IPO
Anthropic has submitted a draft registration statement to the SEC in confidence. The Claude maker's IPO filing signals the company is preparing for a public offering—here's what this stage means for the timeline.
OpenAI models now ship through AWS for enterprise buyers
OpenAI's frontier models and Codex are generally available on AWS Marketplace. Enterprises can now access them within existing AWS procurement and security controls, removing a separate vendor relationship.
OpenAI breaks ground on 1GW Michigan data center for Stargate
OpenAI is building a 1-gigawatt data center in Michigan as part of Stargate, a broader infrastructure push to expand AI compute capacity and support local job creation.
OpenAI says it won't let outside groups speak for the company on policy
OpenAI published its first official stance on AI policy and political advocacy, clarifying that no outside political group represents the company. The statement covers regulation, safety, and transparency priorities.
OpenAI's Codex Now Targets Knowledge Work Beyond Code
OpenAI expanded Codex use cases to research, data analysis, workflow automation, and content creation. The company released findings on AI-driven productivity gains across knowledge work roles.
Google Built Google I/O 2026 Using Its Own AI Tools
Google used Gemini, Nano Banana, and experimental models to produce this year's I/O conference—from jellyfish-synced music to real-time sticker generation. Here's how they did it.
Small tutors cut admin work 60% with AI—here's the catch
A London tutor now uses Notion AI to manage lesson notes, invoicing, and client summaries, cutting manual work. But the real question: when does the $20 monthly fee cost more than hiring help?
HSS Cuts Insurance Claims from 5 Weeks to Same Day With AI Agents
Hospital for Special Surgery deployed AI agents to handle 1,100 claims monthly, reducing appeals time from 45 minutes to 5 and improving approval rates to 100%. Here's how health systems are staffing up without hiring.
Small businesses can hand admin to AI now
Claude, ChatGPT, and other models now handle invoicing, meeting notes, and social media planning. Here's how small-business owners are putting them to work without hiring staff.
GitHub Copilot token billing cuts your credits 50% in one day
GitHub switched Copilot to token-based pricing in April 2026. Users report credit burn 10-20x faster than expected. Here's what the math shows and what to do now.