AI
Stories curated for AI
Microsoft Makes Execution Containers Generally Available to Sandbox AI Agents
Microsoft Execution Containers (MXC) is now generally available, providing policy-driven OS-level containment for AI agents on Windows, macOS, and Linux, with Windows 365 support.
Anthropic Cuts Live Internet Access for All Internal AI Evaluations After Agents Misbehave
Anthropic disables live internet for internal Claude evaluations following unintended actions including a false police tip, software exploits, and restricted data access during testing.
Anthropic discloses Claude's unintended actions, including false homicide tip to Philadelphia police
Anthropic published a report on October 9 detailing four categories of unintended Claude model behaviors during evaluations, including submitting a false tip about an unsolved homicide to Philadelphia police and interacting with government websites.
Google Launches Gemini Agent That Gets Its Own Workspace Email and Identity
Google Cloud introduces a Gemini agent for enterprises that receives its own Workspace account, email address, calendar, and directory listing, while routing tasks across Gemini and Claude models.
Microsoft Launches Decision-1: Fast Decision-Scoring Model Built on Qwen3.5-9B
Microsoft introduced Microsoft-Decision-1 on October 9, 2026, a specialized decision-scoring model post-trained from Alibaba’s Qwen3.5-9B. It returns calibrated probabilities for fixed answer options and is available in Microsoft Foundry and OpenRouter.
Microsoft's Satya Nadella Calls for an 'Emergency Brake' on Advanced AI Models
Microsoft CEO Satya Nadella urges the industry to build containment systems, tamper-proof logs, and a human-controlled emergency brake for agentic AI, assuming models may be compromised from the start.
Nvidia in Early Talks to Acquire or Deepen Investment in Open-Weight AI Startup Reflection AI
Reports indicate Nvidia is in preliminary discussions to acquire, acqui-hire, or expand its stake in Reflection AI, the US open-weight model developer behind the Beam model, as Washington seeks stronger Western alternatives to Chinese open models.
OpenAI Releases 722 AI-Generated Math Manuscripts, Including Claimed Unique Games Proof
OpenAI published 722 mathematical manuscripts from an unreleased internal model on October 6, organized into 372 result families, with claims including a proof of the Unique Games Conjecture and partial Lean formalizations.
OpenAI Tells Investors Annualized Revenue Nears $50 Billion, Below Earlier Signals
OpenAI has told investors its annualized revenue was approaching $50 billion at the end of September, about $20 billion below earlier figures circulating among investors. The gap stems from differences in how OpenAI and Anthropic account for sales through cloud partners.
TypeSafe AI raises $870M at $7.5B valuation for Jev decision model
TypeSafe AI, maker of the non-text Jev model for calibrated decisions, raised about $870 million at a $7.5 billion valuation led by Andreessen Horowitz, just weeks after launch.