Microsoft Copilot Harness Comes to Foundry: What It Costs
Microsoft's Copilot harness is in public preview in Foundry Agent Service. What it does, where it runs, its limits, and the sandbox prices from December 1.

Microsoft's Copilot harness is in public preview in Foundry Agent Service. What it does, where it runs, its limits, and the sandbox prices from December 1.

Google Sheets now shows the web sources behind each AI-filled cell, and Fill with Gemini takes up to 3,000 cells per request. Who gets it, and the limits.

Marvell is in advanced talks to buy optics maker Lumentum, valued at about $83 billion, Bloomberg and Reuters report. What each company does and what is open.

Microsoft's 10-Q shows $10.0 billion in revenue from OpenAI deals in the September quarter, nearly 11% of its $92.3 billion total, plus $7.2 billion owed.

Texas sued six OpenAI entities and Sam Altman in Collin County, seeking a ChatGPT ban for minors and court-approved review of new models. No court has ruled.

Google's Connected Apps Platform preview lets developers bring their services into the Gemini app and Search through MCP servers or Android AppFunctions.

New York's new law bans 12 unsafe AI companion features for users under 18, from bots that claim to be human to flattery. It takes effect January 1, 2027.

AssemblyAI-TTS-1 is AssemblyAI's first text-to-speech model, built for phone voice agents: $15 per 1M characters, 90 ms to first audio, six languages.

Anthropic's Oct. 15 letter to Gov. Newsom backs court orders to suspend risky AI models and says a kill switch alone could give false reassurance.

Claude Enterprise now has smart reports, in-app surveys and pooled group budgets. What each tool shows admins, the privacy limits and who can use it.

Higgsfield Cinema Studio 4.0 makes 30-second AI video scenes from up to 50 references. What changed from 3.5, what clips cost in credits, and API limits.

GitHub Agentic Workflows is now GA: Markdown files that run Copilot, Claude Code or Codex in GitHub Actions, with read-only agents and separate write jobs.

FuriosaAI raised a $620M Series D at a $2.9B valuation. Its RNGD inference card is in mass production, and a 2nm Gen 3 chip is in development with Broadcom.

NVIDIA's open-weight Nemotron 3.5 Transcribe transcribes 40 languages, labels speakers and translates. NVIDIA also opened Granary v2, the data behind it.

OpenAI and ServiceNow expanded their partnership: a ServiceNow MCP plugin for ChatGPT is out now, and security, voice and Codex features are due in 2027.

GitHub Copilot now generates and edits images in chat with OpenAI's GPT Image 2.5 Flare. Which plans get it, how AI credits are billed, and admin controls.

Artificial Analysis rebuilt its Intelligence Index so 60% runs on private or held-out tests. Gemini 4 Carbon now leads at 59, and the open-weight gap grew to 12.

Malaysian customs seized 30 servers with advanced AI chips at KLIA, valued at RM49.5 million (US$12.1 million). Officials call it the fifth such case this year.

Google DeepMind's Gemini 4 Carbon goes first to a subset of cyber defenders, then Fairwind partners. Google's benchmark and safety claims, and the $2/$10 price.

Rime's Prism is a speech-to-speech model for voice agents that writes its reply as text before it speaks. How it works, Rime's numbers and the price.
