Meta Compute Would Turn AI Oversupply Into a Cloud Business
Meta is reportedly developing a cloud infrastructure business that would sell AI compute and hosted model access. The plan is not final, but it shows how Big Tech’s AI data-center spending is starting to look like a market of its own.
X’s Hosted MCP Turns Social Search Into an AI Agent Tool
X has launched hosted MCP servers that let Claude, Cursor, Grok Build, VS Code, and other compatible AI tools call the X API and search X developer docs. The useful shift is not autonomous posting; it is lower-friction access to real-time social data, trends, bookmarks, and platform documentation inside agent workflows.
Verkada and NVIDIA Push Physical AI Deeper Into Security Cameras
Verkada says NVIDIA is now both an investor and technical collaborator as it scales physical AI across more than 2.4 million devices. The deal turns enterprise security cameras into a clearer test case for AI video search, synthetic training data, and governance around real-world monitoring.
Gemini Spark on Mac Turns Desktop Files Into AI Agent Territory
Google is bringing Gemini Spark to the Gemini app for macOS in beta for U.S. Google AI Ultra subscribers. The update gives Spark permission-based access to local files, connected apps, real-time tracking, and soon remote task control from a phone.
California’s Streaming Ad Volume Law Puts Platforms on Audio Watch
California's SB 576 takes effect today, extending TV-style commercial loudness rules to ad-supported streaming services. The law is simple for viewers, but it creates a real compliance test for platforms, ad tech vendors, FAST channels, and device playback chains.
Meta One Puts AI Glasses’ Conversation Focus Behind a Usage Meter
Meta’s AI glasses now have monthly usage limits for Conversation Focus, with free users capped at three hours and Meta One Premium subscribers capped at 15. The change turns a useful wearable audio feature into an early test of how far consumer AI hardware subscriptions can go.
Claude Fable 5 Returns With a New Test for AI Jailbreak Rules
Anthropic is restoring Claude Fable 5 after U.S. export controls on Fable 5 and Mythos 5 were lifted. The redeployment brings a new cyber-safety classifier, fallback handling for blocked requests, and a proposed industry framework for scoring AI jailbreak severity.
BlueHammer Ransomware Flag Puts Microsoft Defender Patching Back on the Clock
CISA has updated the Microsoft Defender BlueHammer flaw, CVE-2026-33825, to mark it as used in ransomware campaigns. The flaw was patched in April, but the new flag gives Windows teams a fresh reason to verify Defender updates, endpoint telemetry, and local privilege escalation controls.
Microsoft Defender Starts Watching Local AI Agents on Developer Machines
Microsoft Defender now discovers local AI agents and MCP server configurations across managed endpoints, while preview runtime protection can audit or block prompt-injection attempts in Claude Code and GitHub Copilot CLI before risky tool actions execute.
Claude Sonnet 5 Makes Agentic AI Cheaper to Run
Anthropic launched Claude Sonnet 5 with lower launch pricing, stronger agentic behavior, Claude Code support, and broad availability across Claude plans. For developers, the useful question is not whether it is the flashiest Claude model, but whether its cost, context window, and migration changes make long-running agents easier to put into production.