👀 OpenAI just put its coding agent on your phone.
PLUS: Anthropic overtakes OpenAI in US business spending for first time
OpenAI just put its coding agent on your phone.
OpenAI rolled out Codex in the ChatGPT mobile app as a preview for all plans on iOS and Android on Thursday, including its entry level Free and Go tiers in accessible regions. For the first time, developers are able to track, manage, and send coding tasks from their phones.
How it works:
Your phone does not run Codex yet. It connects to Codex. Anything that a developer has already setup in the environment, a Mac, a devbox or any remote server, remains untouched. ChatGPT acts as a portal into that world, providing live streams of threads, outputs, screenshots of results, test runs and approvals. Files, credentials and permissions remain on the Codex machine. A developer can begin a task in the office, leave for the day and then resume on their phone when the agent hits a decision point and requires guidance.
A key restriction at the moment of launch: the mobile app in its initial form only interfaces with Cordex's macOS desktop app. Windows support (date TBD, though it's probably coming as well)
Beyond the feature, why this is important
Codex launched in April 2025. There was also a stand-alone Mac app which launched in February 2026. From the time they have launched, mobile access has been a clear miss in the product and something developers were hoping to see.
Anthropic moved first here. It has been approximately two and a half months since Claude Code launched Remote Control on February 25, 2026 (for Max and Pro subscribers), OpenAI is filling that gap today and it's making one big change in the process: Codex mobile costs nothing to use across every plan starting now, while Claude Code's Remote Control launched natively as a premium feature for Max subscribers ($100 to $200/month) with Pro rollout promised but limited at first.
Codex currently boasts more than 4 million weekly users, according to OpenAI. What the company sent out in terms of a mobile update was not an only Thursday. Hooks made its way to general availability the same day, alongside CI-pipeline and release-workflow programmatic access tokens, and HIPAA-compliant Codex for qualifying Enterprise workspaces in local environments. Maybe the mobile consumer-facing headline will matter less long-run than the enterprise plumbing.
The bigger picture
OpenAI in March confirmed it was developing a single desktop app that rolls ChatGPT, Codex, and its Atlas web browser into one seamless environment. The vision is an AI that lives with you similar to your entire working life rather than one living in a tab you open. Claude Code, Cowork, and Remote Control from Anthropic are taking you to the same place. Perplexity (Personal Computer for mac mini), a desktop version of the personal assistant: now has always-on background agent functionality. Each big lab is coming to the same idea from different angles.
Developer market is the primary battleground, because developers are the deepest users and builders of what eventually gets built by everyone else. By February 2026, its annualised revenue run rate was $2.5 billion, making it Anthropic's fastest-growing product. OpenAI has withheld commensurate revenue statistics for Codex in general.
OpenAI's update on Thursday brings the company up to speed with Anthropic's capabilities back in February for mobile access, but it expands much further by making availability possible at a free-tier level. The coming months will reveal whether that is enough of a reason for the four million weekly Codex users to remain and if Claude Code users will give it a second glance.
POLL: We cover AI every week inside Cryptopolitan Daily. Would you want a dedicated daily AI newsletter from us?
Yes, add me to the AI newsletter list
Maybe, keep sending the weekly for now
No, the weekly coverage is enough for me
I don't follow AI closely
🤖 AI Watch
Figma stock rallied 13% after beating Q1 earnings, but analysts flagged the Anthropic-Trump standoff as a growing risk. Figma's design platform has deep integration with Anthropic's Claude tools, including the recently launched Code to Canvas partnership. The Pentagon's supply chain risk designation against Anthropic, and the broader White House friction, creates a vendor dependency risk that analysts are now putting explicitly in filings. A company whose AI roadmap runs through a partner that the US government is simultaneously suing and deploying is carrying a risk that did not exist six months ago.
Elon Musk missed closing arguments in the OpenAI trial as the jury began deliberations. Musk was travelling with the presidential delegation to Beijing when his own lawyers were making their final case to a San Francisco jury. The trial has been running four weeks. The jury is now deciding whether OpenAI betrayed its nonprofit founding mission by removing its profit cap and raising $122 billion. Musk is seeking $134 billion in damages from OpenAI and Microsoft. His absence from the courtroom on the most important day of the case is the kind of detail that tends to follow a story permanently.
SpaceX is filing its public IPO prospectus within the next couple of days. The S-1 going public means retail investors can read the full financials, risk disclosures, and governance structure for the first time. The company is targeting a $1.75 trillion valuation and aims to raise $75 billion. The xAI CSAM regulatory exposure flagged in the risk section, the super-voting share structure that gives Musk 79% voting control, and the $603 million Bitcoin position will all be visible in the public document. The roadshow is expected to begin the week of June 8.
Bithumb warned users as AI scams stole millions in crypto. South Korea's largest crypto exchange issued a formal user warning after AI-powered social engineering attacks drained millions from customer accounts. The scams use synthetic voice cloning and deepfake video to impersonate exchange staff, financial advisors, and even family members. The same capability gap that Anthropic's Mythos exposed in software security is showing up in social engineering: AI is lowering the cost and skill threshold for sophisticated fraud to near zero.
Are you watching
More American companies now pay for Anthropic’s Claude than for OpenAI’s ChatGPT. That’s according to expense data from Ramp, a fintech platform tracking over 50,000 U.S. businesses.
Meme of the day
Join the Conversation!
We'd love to hear your thoughts and comments. Join our community and stay updated with the latest trends and discussions in crypto.
© 2026 Cryptopolitan
75 East 3rd Street, Suite 7 Sheridan, Wyoming 82801, United States
Send this story to anyone — or drop the embed into a blog post, Substack, Notion page. Every play sends rev-share back to storyflo · listener sends.
70% to the publisher · 20% to whoever forwarded this · 10% keeps Storyflo running.
Loading comments...