Answer-First Lead
Anthropic is quietly eating OpenAI’s enterprise lunch on Ramp’s platform. Google DeepMind is rethinking the mouse pointer — the most durable user interface in computing history — for an AI-native world. Android is deploying AI to detect AI-powered scams before they reach your phone. And Google is now asking software engineering candidates to use AI during their interviews — because pretending they won’t is pointless. Let’s get into it.
📰 Stories
1. 📊 Anthropic Quietly Overtakes OpenAI in Enterprise Adoption on Ramp
The story: Data from Ramp — the corporate spend management platform that also tracks what AI tools companies are paying for — shows Anthropic’s Claude has overtaken OpenAI’s ChatGPT in enterprise customer adoption on its platform. Ramp processes payments for over 30,000 businesses, giving it a unique window into real-world enterprise AI spend.
The data, reported by Fortune on May 15, shows Claude being adopted by more net-new enterprise customers month-over-month, particularly in regulated industries like finance, legal, and healthcare. Companies cite Claude’s larger context window, more predictable pricing, and stronger safety guardrails as reasons for the switch.
Why it matters: API revenue data is noisy. What Ramp shows is actual payments flowing from enterprise customers to AI vendors — and Anthropic is winning on net-new logos. OpenAI still dominates in consumer mindshare and total revenue, but the enterprise narrative is shifting. If Claude becomes the default AI for regulated industries, the competitive landscape looks very different in 12 months. The question is whether OpenAI’s Deployment Company (see News digest) is specifically designed to counter this trend.
2. 🖱️ Google DeepMind Reimagines the Mouse Pointer for the AI Era
The story: Google DeepMind published research on May 12 proposing a fundamental redesign of the mouse pointer — a UI element that has been essentially unchanged for over 50 years. The new design adapts the cursor to show not just where it is, but what it can interact with, what the AI understands about the context, and what actions are available.
The research explores “semantic pointers” that change shape based on the type of content being hovered — expanding to show AI-suggested actions, contracting to indicate non-interactive regions, and colour-shifting to signal confidence levels. Think of it as the cursor becoming a real-time AI interface rather than just a pointing device.
Why it matters: The mouse pointer has been frozen in amber since the Xerox Alto. DeepMind is asking a genuinely interesting question: if everything on your screen is AI-augmented, why should your cursor behave the same way it did in 1973? This could be the first step toward a genuinely new interaction paradigm — or it could be a research curiosity that never ships. Either way, someone needed to question the 50-year-old icon.
3. 📱 Android Pushes New Scam, Theft, and AI Protections in 2026 Update Wave
The story: Google’s May 2026 Android security update, detailed by Help Net Security, introduces new scam detection powered by on-device AI that analyses call and message patterns in real time. The system can detect voice phishing attempts, SMS social engineering, and — critically — AI-generated voice clones used in scam calls.
The update also includes enhanced theft protection (remote locking even if the phone is offline) and expanded app permission controls specifically for AI-powered apps that request runtime data access.
Why it matters: AI-powered scams are the fastest-growing category of consumer fraud. Voice cloning alone is enabling a wave of “grandparent scams” that are almost impossible to distinguish from genuine distress calls. Android putting on-device AI scam detection into the OS — where it works without needing another app — is a meaningful step. The challenge is the asymmetry: scammers only need to succeed once, the defender needs to catch everything.
4. 💻 Google Lets Software Engineering Candidates Use AI in Interviews
The story: Google is reportedly piloting a new interview format for software engineering roles where candidates are told explicitly that they can use AI tools during the technical assessment. The company wants to evaluate how candidates use AI as a productivity tool — not whether they can write code without it.
This is a significant shift from the traditional “closed-book, no tools” technical interview model that has defined Big Tech hiring for two decades. Google is effectively saying: in the real world, you’ll have AI assistance. We want to see how good you are with it.
Why it matters: Every company still hiring engineers is wrestling with the same question: what is a software engineer worth when AI can write 80% of the code? Google’s answer — “let’s test their ability to use AI effectively” — is honest about the current reality. But it creates a new arms race in interview prep: now you need to practice AI-assisted problem-solving, not just algorithmic thinking. The signal to noise ratio of these interviews is about to change dramatically.
5. 🏭 Nvidia CEO Says Software Engineers Are Busier Than Ever
The story: Nvidia CEO Jensen Huang dismissed concerns that AI is destroying software engineering jobs in a May 16 interview, arguing that “software engineers are busier than ever.” Huang’s position is that AI increases the demand for software by making it more capable and accessible — which in turn creates more engineering work, not less.
He acknowledged that the nature of the work is changing — engineers spend less time writing boilerplate and more time on architecture, integration, and creative problem-solving. But he pushed back on the narrative that AI will reduce total engineering headcount.
Why it matters: Jensen Huang has a vested interest in this narrative — Nvidia sells the chips that run AI. But his argument isn’t wrong: every major productivity tool in history created more knowledge work, not less. The spreadsheet didn’t eliminate accountants; it created more of them. The question is whether AI follows the same pattern — or whether this time is genuinely different because AI doesn’t just augment, it replaces cognitive work directly. The answer probably depends on which decade you’re looking at.
6. 🇸🇬 Singapore to Upskill 40,000 Tech Workers in AI
The story: Singapore’s Infocomm Media Development Authority (IMDA) announced a national initiative to upskill 40,000 tech workers in AI by 2028, as reported by Human Resources Director. The programme covers foundational AI literacy, hands-on training with AI development tools, and specialised tracks for AI ethics and governance.
Singapore is positioning itself as Asia’s AI hub, competing with cities like Tokyo, Seoul, and Bangalore for talent and investment. The government is subsidising training costs and partnering with major tech companies to provide curriculum and certification pathways.
Why it matters: Singapore is one of the few countries treating AI workforce transformation as a national priority with funded programmes attached. Compare this to most Western nations, where AI upskilling is left to individual companies, bootcamps, or self-directed learning. The Singapore model — government-led, industry-partnered, with measurable targets — is worth watching. If they hit 40,000 skilled workers in two years, it sets a benchmark for what intentional policy can achieve.
🔍 THE BOTTOM LINE
Anthropic is quietly winning the enterprise AI race where it counts — in real payments. Google DeepMind is daring to redesign the most durable interface in computing. Android is fighting AI scams with AI defences. Google’s interview process just admitted that pretending AI doesn’t exist is pointless. Jensen Huang says the “AI kills jobs” narrative is wrong. And Singapore is showing what deliberate workforce policy looks like. The theme this week: adaptation isn’t optional anymore.