🤖 OpenAI ships Codex to ChatGPT iOS app 🤖
Developers can now access their Codex coding agent remotely from any iPhone.

OpenAI has rolled out its Codex coding agent to the ChatGPT iOS mobile app, letting developers track and manage long-running AI tasks, review code changes, handle approvals, interact with plugins, and start new tasks — all from their phone. The system connects to ongoing work on laptops, devboxes, or remote environments through a "secure relay layer" that avoids exposing the user's machine to the open internet while syncing across ChatGPT instances. This is a key distribution move for portable agentic coding. The rollout marks OpenAI's push to position Codex as an always-available development companion, and the launch was the headline feature across multiple developer-focused outlets.

🔬 Nvidia partners with AlphaGo creator on RL training 🔬
Nvidia taps DeepMind's David Silver to build reinforcement learning infrastructure at scale.

Nvidia has announced a partnership with Ineffable Intelligence, a UK-based startup founded by UCL professor David Silver — the researcher behind AlphaGo and AlphaZero. The collaboration will build training pipelines for reinforcement learning agents, with early work targeting Nvidia's next-generation Vera Rubin hardware. This signals competitive pressure as Nvidia moves beyond GPU supply to shape frontier training methods. The deal is notable for its European dimension and represents a high-profile bet on RL-driven approaches to superintelligent systems, with implications for both the AI research landscape and the future of large-scale agent training.

💰 Anthropic adds agent credit caps to Claude plans 💰
New pricing policy separates agentic usage from standard Claude subscription limits.

Starting June 15, Anthropic will separate Claude's agent usage into a dedicated monthly credit pool. Pro users receive $20/month in agentic credits, Max 5x users get $100, and Max 20x users get $200 — credits reset each billing cycle and do not roll over. The Agent SDK and claude -p will no longer draw from standard subscription quotas, reversing Anthropic's controversial April ban on third-party agents. However, power users have pushed back, calling the caps too restrictive. This pricing signals how frontier labs are trying to balance surging agent demand with sustainable economics, a challenge that matters as agentic workflows become the default for developers.

INTERESTING TO KNOW

⚖️ OpenAI eyes legal action against Apple over Siri ⚖️

OpenAI is reportedly preparing legal action against Apple over how ChatGPT was integrated into Siri following their 2024 partnership. The rollout underperformed expectations — OpenAI anticipated "billions" in paid signups, but internal data shows users prefer the standalone ChatGPT app over the Siri-embedded experience. Tensions have deepened as Apple plans to open Siri to rivals including Claude and Gemini in iOS 27, while OpenAI has been poaching Apple hardware engineers. This dispute could reshape future AI–device distribution deals across the industry.

🛠️ xAI launches Grok Build CLI coding agent 🛠️

xAI has released Grok Build, a CLI-based agentic coding tool now in early beta with access limited to SuperGrok Heavy subscribers. It supports AGENTS.md, plugins, hooks, MCP servers, and subagent capabilities out of the box, with deep worktree integrations and a headless mode for automation. The launch positions Grok Build as a direct competitor to OpenAI Codex CLI and Anthropic Claude Code, targeting power-user developer workflows with a composability-first approach.