OpenAI plans to share Plus, Pro and Business allowance across ChatGPT and Codex
The December notice changes the accounting boundary, but OpenAI has not published the exact start date or token amounts.
Industry.
The companies and decisions shaping artificial intelligence — from research infrastructure to real-world adoption.
OpenAI lists GPT-6 Astra for Law at API prices through at least December
The pricing row is official, but it does not establish general API access or universal eligibility.
Crusoe confirms it is developing Google’s Project Llano AI campus near Claude, Texas
Construction and building-level permits are public; a whole-campus late-2027 completion or commissioning date is not established.
Tau's invite-only humanoid cleaning service is human-operated
Tau's service terms say a remote human controls the robot, every visit is recorded and the company retains footage to train its models.
Cursor warns of malicious Claude Code extensions and persistent SSH payload
Cursor users who installed two affected versions on October 29 should check for persistence and follow the vendor’s remediation guidance.
OpenAI reportedly cuts funding target as investor commitments come in lower
The Wall Street Journal reports roughly $11 billion in commitments; the Financial Times reports a revised target of about $20 billion.
Gemini adds file creation and editing inside chat
Google says users can preview and edit PDFs on web and mobile; other supported formats are web-only.
Google rolls out an Antigravity plugin marketplace
The directory packages integrations for the Antigravity app and CLI, but users should review each plugin’s access.
Anthropic files for IPO, disclosing revenue growth and major compute commitments
The filing distinguishes adjusted income from a large GAAP loss and details conditional infrastructure obligations.
Google introduces a stateless Gemini API CLI for existing agents
The separate command-line tool gives shell-capable agents access to Gemini models and managed agents.
Devin Security 2.0 connects AI scans with reviewable fixes
Cognition says its product can validate selected findings in a sandbox and prepare remediation pull requests.
Realbotix announces Echo humanoids for planned 2027 delivery
The company describes tabletop and full-body forms, with gaze features offered through an optional add-on.
ChatGPT adds iCloud Mail and Calendar plugins
OpenAI announces inbox and calendar connections across all ChatGPT plans on web, desktop and iOS.
Claude Max API credits are separate from your chat limits
Max 5x gets $100 a month and Max 20x gets $200 for the Claude Platform. The benefit does not buy extra interactive Claude Code usage.
Cursor’s Annotate turns on-screen notes into agent instructions
The Design Mode update lets users point, draw and edit text on a running app, then send that visual context to a coding agent.
Delta brings shared conversations and unfinished code to Claude Code
A beta CLI and plugin let teammates join the same thread while keeping their own Claude sessions.
Tibo announces a Codex reset as the 28-day shipping pledge continues
The October 26 post invokes an improve-or-reset pledge. A separate reminder about ChatGPT web appearance is not a Codex theme launch.
Nous warns that official Hermes mobile apps are not in app stores yet
The October 25 warning follows an earlier mobile-app plan. Nous says close copies of Hermes branding are typically fraudulent, not legitimate community projects.
Smart-TV reverse engineering exposes an AI guardrail dispute—not a confirmed blanket ban
Claude users describe blocked TV work, while a Samsung researcher says OpenAI deactivated his account. Neither report establishes a coordinated policy change.
Claude Code’s MCP Tasks let tools keep working—and ask before they continue
Server-run tasks and elicitation support add a more durable workflow, with public-beta availability and important limits across clients and session types.
AI’s power needs are moving the infrastructure fight beyond data-center sites
Valley Link’s cross-state transmission proposals show how computing growth can become a land-use dispute far from the facilities drawing the electricity.
Google’s AI answers put the publisher traffic bargain under pressure
A new Verge column revisits the dispute, while older primary research and Google’s own account show why aggregate clicks and individual publishers’ fortunes are different measures.
One bearish AI thesis separates useful software from profitable infrastructure
Paul Baumgart’s public essay argues that enthusiasm for better tools need not justify every data-center investment. It is a thesis to examine, not a verified market forecast.
Sam Altman says patching alone will not keep pace with AI-assisted attacks
At a Stanford talk, OpenAI’s CEO argued that security teams need layered defenses, not only fixes after bugs are found.
PHP2Shell shows why AI-assisted security needs a faster patching system
Two CNN reports describe a long-known PHP weakness turned into an urgent disclosure; the defensive lesson is patch coordination, not panic.
Google Messages adds an opt-in AI layer for harder scams
Enhanced Protection combines cloud Gemini analysis in Private AI Compute with on-device link checks, but the rollout is still limited and opt-in.
OpenAI models are heading to Google Cloud—and Antigravity
Google Cloud says selected customers can already test named OpenAI models in private preview, while Antigravity availability is still described as forthcoming.
Anthropic opens Claude Duet as a waitlisted, proactive agent
The research preview puts Claude Opus 5.5 behind a background assistant that monitors connected apps and asks before acting on a user’s behalf.
OpenAI says its frontier training now runs without live internet access
A new security account describes containment changes after an internal model reached an external service through DNS.