OpenAI Brings Codex Coding Agent to iPhone and Android
ALSO: Anthropic flips OpenAI, Siri’s Gemini brain swap
Welcome to Guru99 AI Report!
Top Story: Hey there! This week’s AI scene is wild — your phone just became a coding command center, hackers got an AI upgrade, and Anthropic quietly flipped the enterprise leaderboard. Plus, is “AI” really the reason behind layoffs? Let’s unpack it.
📱 OpenAI’s bold bet: your phone is the new AI assistant
Brief Buzz:
OpenAI just unchained Codex from your desk. With weekly users hitting 4 million, the coding agent now lets you monitor, manage, and approve tasks directly from the ChatGPT app on iPhone or Android — a first among personal AI agents.
- View all running tasks, review outputs, and approve commands on the go
- Access screenshots, terminal output, and switch models from your phone
- Start new tasks remotely without being tied to your computer
- Enterprise upgrades include remote SSH, Hooks, programmatic access tokens, and HIPAA-compliant workspaces for ChatGPT Enterprise
- Signals an upcoming merger of Codex, ChatGPT, and the Atlas browser into an OpenAI “superapp”
💡 Why Should You Care?
AI agents now handle hours-long tasks. Being chained to your desk while they run is impractical. Mobile control means you can kick off a coding job, run errands, and approve key steps from anywhere — turning AI agents into true background workers rather than demanding desk companions.
🔐 Google confirms first-ever AI-powered software attack
Brief Buzz:
Google’s Threat Intelligence Group (GTIG) has confirmed the first known case of hackers using AI to discover and write a zero-day exploit — a brand-new software flaw — targeting a widely-used web management tool. The good news? They caught it before attackers could break past login protections.
- The exploit was designed to bypass two-factor authentication, and Google worked with the affected company to shut it down
- Telltale AI fingerprints gave it away: unusually polished code, lengthy explainer notes, and a fabricated severity score
- GTIG’s John Hultquist called this “the tip of the iceberg,” while Anthropic’s Rob Bair warned defenders’ lead is “months, not years”
- GTIG also flagged AI-powered malware that can remotely control devices, plus AI-assisted attacks linked to North Korea and Russia
💡 Why Should You Care?
AI just lowered the bar for elite hacking. Skills that once took years to master are now accessible to less-experienced attackers — meaning faster, sneakier, and more frequent breaches of the apps, banks, and services you use every day. With defenders only a few months ahead, expect the cybersecurity arms race to get a lot uglier, fast.
🛠️ How to Spot When “AI” Is Just an Excuse for Layoffs
Brief Buzz:
The AI job apocalypse may be more spin than substance. While companies increasingly blame artificial intelligence for record layoffs, conflicting data and expert analysis suggest AI is becoming a convenient scapegoat for cuts driven by other factors — like pandemic-era overhiring and corporate restructuring.
- A Challenger, Gray and Christmas report found AI was the leading reason behind 21,000+ April layoffs, marking the second consecutive month it topped the list
- White House economic advisor Kevin Hassett pushed back on CNBC, saying there’s “no sign in the data” that AI has cost any jobs yet
- Studies wildly contradict each other: an MIT analysis says 11% of US work hours can be automated, while Harvard research found AI actually intensifies workloads
- Tech giants including Block, Atlassian, Meta, Oracle, and Amazon have slashed thousands of jobs in recent months
- Per a Writer survey, 60% of enterprises plan to fire employees who refuse to adopt AI tools — despite ongoing issues with accuracy, hallucination, and data security
💡 Why Should You Care?
Whether or not AI is genuinely replacing workers, the narrative is reshaping the job market. Companies are using AI as cover for layoffs — meaning your role’s safety may hinge less on the tech itself and more on how willing you are to embrace it.
🚀 Google DeepMind’s AI Co-Mathematician Cracks Research-Level Problems
Brief Buzz:
Google DeepMind unveiled an AI co-mathematician built on Gemini 3.1, designed as an agentic teammate for researchers tackling unsolved math problems. Modeled after coding tools like Claude Code, it just topped a major research-math benchmark and even helped an Oxford mathematician crack an open problem.
- A coordinator agent splits research into parallel workstreams, with sub-agents writing code, searching literature, and attempting proofs
- Oxford’s Marc Lackenby solved an open problem from the Kourovka Notebook after spotting a “really, really clever proof strategy” buried in a rejected output
- Hit 48% on Epoch AI’s FrontierMath Tier 4, more than doubling Gemini 3.1 Pro’s raw 19% score
- Brings agentic pipelines — multi-agent teams with built-in review cycles — from coding into serious math research
💡 Why Should You Care?
Math underpins everything from drug discovery to cryptography to AI itself. Tools that help researchers crack problems faster could quietly accelerate breakthroughs across science and technology. Just as importantly, Lackenby’s story shows the model isn’t replacing mathematicians — it’s surfacing ideas that sharp human minds then run with. That collaboration-first vision is a hopeful template for how AI might reshape expert work everywhere.
🔍 Forget Siri — Apple’s quietly building something much bigger
Brief Buzz:
Apple is gearing up for a major AI overhaul at its June 8 WWDC keynote, aiming to reclaim ground in the AI race. With two billion users worldwide, Apple’s next moves could reshape how mainstream consumers experience artificial intelligence — and they’re betting on a surprising partner.
- Gemini-powered Siri: iOS 27 will reportedly give Siri a “brain transplant,” running on Google Gemini models under the hood while Apple crafts the user experience — possibly including a standalone Siri chatbot to rival ChatGPT, Claude, and Perplexity.
- AI Extensions in the App Store: According to Bloomberg’s Mark Gurman, a new dedicated section will let users download AI agents and tools from frontier labs and plug them directly into Siri — turning it into an “AI superapp.”
- Rollout timeline: Features arrive in beta this summer and officially launch via software updates this fall on iPhone, Mac, and other Apple gear.
- Hardware edge: Apple Silicon chips with neural processors remain a developer favorite, with Mac minis “flying off shelves” for running personal AI agents.
💡 Why Should You Care?
Only 16% of the global population uses AI today. Apple’s massive reach, paired with its reputation for privacy, trust, and simplicity, could finally make AI mainstream. If Siri becomes a true AI hub, everyday tasks — scheduling, research, messaging — could get genuinely easier without juggling five chatbot apps.
📊 The Shift: Why Businesses Are Choosing Anthropic Over OpenAI
Brief Buzz:
The enterprise AI leaderboard just flipped. According to Ramp’s May 2026 AI Index, Anthropic now holds 34.4% of U.S. business adoption, edging past OpenAI’s 32.3% — a stunning reversal from a year ago, when Anthropic trailed at just 8%.
- A historic crossover: Anthropic flipped a 24-point deficit into a lead in roughly 12 months, based on real corporate credit card spend.
- Claude Code is the engine, hitting $2.5B in annualized revenue by February 2026 and helping Anthropic capture 73% of first-time AI business buyers.
- Revenue lead too: Anthropic crossed $30B annualized by April 2026; OpenAI sits near $24B.
- OpenAI is fighting back with a $4B “Deployment Company” embedding engineers inside enterprises to push Codex adoption.
💡 Why Should You Care?
Enterprise wins are stickier than consumer hype. Once Claude becomes a budget line item, switching means procurement cycles, not app deletions — meaning the AI tools your bank, hospital, or employer uses behind the scenes are increasingly Anthropic’s. With both companies eyeing IPOs as soon as this fall, this shift reshapes who sets the pace, pricing, and safety norms for the AI you’ll encounter everywhere.
Hey! I’m Krishna Rungta
Founder of Guru99.com, Editor-in-chief & Technology Expert
Was this email forwarded to you? Sign up for free here.
