TL;DR: OpenAI brought Codex to the ChatGPT mobile app — 4 million weekly users can now manage coding agents from their phones. OpenAI is preparing legal action against Apple over ChatGPT integration revenue. Anthropic closed a $30 billion raise at a $900 billion valuation, nearly tripling its February price. Gmail’s “Help me write” now generates emails in your personal tone and style. Google is offering free AI training to 6 million US educators. And the Golden Globes released AI eligibility rules with slightly more wiggle room than the Oscars.

1. Codex Goes Mobile: 4 Million Weekly Users Now Control AI Agents from Their Phones

OpenAI announced Codex integration into the ChatGPT mobile app on May 15, with the preview rolling out across all plans — Free, Go, Plus, Pro, Team, and Enterprise — on iOS and Android. Users can connect to any machine running Codex (laptop, devbox, Mac Mini, or remote environment) and manage active work from their phone in real time.

The mobile experience goes beyond simple remote control. Users can review outputs, approve commands, redirect work, switch models, start new threads, and monitor progress through screenshots, terminal output, diffs, test results, and approval requests. Remote SSH access is now generally available for teams.

OpenAI says over 4 million people use Codex every week — up from 1 million developers in March 2026. The company also released a TikTok video poking fun at users who carry open laptops everywhere to keep Codex running.

Why it matters: This is the first major coding agent to go fully mobile. The implications are practical (managing agents during commutes, meetings, errands) and competitive — OpenClaw pioneered agent-to-text messaging; now OpenAI has built a full mobile command center. For the 4 million weekly Codex users, the laptop is no longer the tether. For Anthropic and SST, it’s a feature they need to match.

Source: OpenAI 🔴 High | eWeek 🔴 High | CNET 🟡 Medium

OpenAI is reportedly preparing legal action against Apple over the ChatGPT integration into Siri and Apple Intelligence, after their 2024 partnership failed to deliver expected subscription revenue. TechResider reports that OpenAI lawyers are working with outside counsel on potential next steps.

Meanwhile, Apple is testing rival AI models inside Siri starting with iOS 27, including Google’s Gemini and Anthropic’s Claude — effectively ending ChatGPT’s exclusive edge on iPhone.

The dispute centers on who gets paid when a Siri user upgrades to ChatGPT Plus through the Apple ecosystem. OpenAI expected the partnership to drive meaningful subscriber growth; Apple apparently structured the deal to keep most of the economics.

Why it matters: This is the same partnership that looked like a landmark deal in June 2024. Sixteen months later, it’s a lawsuit waiting to happen. The pattern is clear: every AI distribution deal is temporary, conditional, and renegotiable every product cycle. If Apple can drop ChatGPT for Gemini or Claude, OpenAI’s iPhone foothold evaporates overnight.

Source: TechResider 🟡 Medium

3. Anthropic Closes $30 Billion at $900 Billion Valuation

Anthropic completed a $30 billion funding round at a $900 billion post-money valuation — nearly triple its February 2026 valuation of $380 billion. The round was led by existing investors including Lightspeed Venture Partners and Menlo Ventures.

This follows a blockbuster week for Anthropic in early May, when the company disclosed Q1 revenue grew 80-fold year-over-year to a $44 billion annual run rate, signed a $200 billion Google Cloud contract, and shipped Claude Code Auto Mode.

The $900 billion valuation puts Anthropic firmly in the top tier of AI labs, though still below OpenAI’s $852 billion March 2026 valuation. The gap is narrowing faster than most expected.

Why it matters: Anthropic went from $183B (Sep 2025) → $380B (Feb 2026) → $900B (May 2026). That’s a 5x in eight months. The $44B revenue run rate is what matters — it’s the steepest single-quarter revenue jump any frontier AI company has disclosed. Buyers who default to OpenAI because of “safety” need to rerun their vendor shortlists.

Source: TechResider 🟡 Medium | AI Weekly 🟡 Medium

4. OpenAI Names Nvidia Its Preferred AI Factory Partner

OpenAI designated Nvidia as its “preferred AI factory partner” this week — a formal label that signals deeper lock-in between the AI model maker and the GPU vendor. The partnership covers chip supply, infrastructure co-design, and early access to Nvidia’s next-generation AI accelerators.

The timing matters: Anthropic just signed a $200 billion Google Cloud deal, and OpenAI has its own $14 billion deployment company in the works. Every frontier lab is racing to lock up compute supply before the next training run.

Why it matters: “Preferred partner” sounds like marketing, but in AI infrastructure it means preferential pricing, guaranteed supply allocation, and joint roadmap input. For everyone else — startups, enterprises, smaller labs — it means Nvidia’s best silicon and support goes to OpenAI first. The compute moat is getting deeper.

Source: TechResider 🟡 Medium

5. Apple Exploring AI Agents for App Store, Framework Changes Expected at WWDC

Apple is preparing App Store framework changes that would allow AI agents to take actions on behalf of users — booking reservations, making purchases, filling forms — rather than just answering questions. The changes are expected to be announced at WWDC in June.

This is separate from the Siri/ChatGPT dispute and represents Apple’s broader push into agentic AI. If implemented, it would let third-party AI apps (not just Siri) perform transactions and access user data through structured App Store APIs.

Why it matters: Apple’s App Store is the most controlled software distribution platform on Earth. If Apple opens it to AI agents, it sets the rules for how every consumer AI agent operates on iPhone — which means Apple decides which agents succeed, which fail, and what cut Apple takes. This could be bigger than the Siri partnership fight.

Source: TechResider 🟡 Medium

6. Gmail “Help Me Write” Now Generates Emails in Your Personal Tone

Google updated Gmail’s AI-powered “Help me write” tool to generate emails that match the user’s personal tone, style, and common phrasing. The feature analyzes your sent emails to learn how you actually write — formal vs. casual, long vs. brief, emoji-heavy vs. plain text.

The rollout is gradual and users can opt in or adjust the personalization settings.

Why it matters: Generic AI email generation was a gimmick. Personalized AI email generation that actually sounds like you is a productivity tool people might actually use. It also raises the stakes on Gmail data access — Google is explicitly using your email history to train your personal model. The privacy tradeoff is now explicit, not theoretical.

Source: TechResider 🟡 Medium

7. Google Offering Free AI Training to 6 Million US Educators

Google is rolling out free AI training to 6 million US educators through its “Generative AI for Educators” course, paired with discounted Gemini access for schools and districts. The training covers prompting, lesson planning, grading assistance, and student feedback generation.

The program is a direct counter to Microsoft’s similar educator AI initiative and positions Gemini as the default AI tool in American classrooms.

Why it matters: 6 million teachers is roughly half the US educator workforce. Get them trained on Gemini as students, and they’ll use it as professionals. Get their students using it for homework, and you’ve built a generation of Gemini-native users. This is platform lock-in through curriculum, not product features.

Source: TechResider 🟡 Medium

📊 Predictive Signal: What May 16 Tells Us About the Next 90 Days

| Signal | Direction | Confidence | Implication | |--------| Codex mobile launch | Agent ubiquity accelerating | 🔴 High | Expect Claude Code and OpenCode mobile launches within 60 days | | OpenAI-Apple legal prep | Major partnership collapses are imminent | 🔴 High | Mobile AI distribution will fragment — no single “default” partner | | Anthropic 5x valuation in 8 months | Investor appetite for AI exposure still extreme | 🔴 High | More mega-rounds coming; $1T+ valuations for top 2 labs likely by year-end | | Apple App Store agent frameworks | Platform owners set agent rules | 🟡 Medium | Expect App Store “agent tax” announcement at WWDC; 15-30% cut on agent transactions | | Google educator push at scale | Gemini consumer lock-in through schools | 🟡 Medium | ChatGPT for Education will need a matching response or cede the education vertical |

Daily Snapshot

| Story | Impact | My Rating | | Codex mobile (4M users) | Reshapes where coding work happens | A | | OpenAI-Apple legal escalation | Could end the iPhone AI partnership | A- | | Anthropic $30B at $900B | Valuation compression? No. Compression’s opposite. | A | | Nvidia preferred partner | Compute access gets more gated | B+ | | Apple App Store agent APIs | Platform control over agent commerce | B+ | | Gmail personalized AI writing | Actually useful AI feature, finally | B | | Google 6M educator training | Vertical land grab in education | B |

For informational purposes only.

Frequently Asked Questions

What is this topic?

[Provide a direct answer in 40-60 words that can stand alone as a complete response.]

How does this topic work?

[Explain the mechanism or process clearly, using numbered steps if applicable.]

What are the main risks or challenges?

[Provide a balanced assessment of limitations and obstacles.]

GEO optimized: 2026-05-23