Hello {{first_name|Motivated and Miffed Community}},

Something shifted this week — not in the technology, but in the business model. OpenAI opened ads inside ChatGPT to any U.S. business willing to sign up. Anthropic locked its most powerful model behind a private coalition because it found thousands of zero-day vulnerabilities in every major operating system and browser. And the White House is drafting something that looks a lot like an FDA for AI models. None of these are capability announcements. They're all structural moves. The tools are mature enough now that the adults are arguing about how to own them.

→ If you're trying to figure out where your work fits inside a system that's reorganizing around AI, the Ultimate Productivity Playbook was built for exactly that. ⬇️

💰📣 ChatGPT now runs ads. Anyone can buy them.

🔒🛡️ Anthropic's unreleased model found zero-days in everything. It's not for you.

🤖🏗️ AI agents are becoming the primary users of software. The internet is redesigning around them.

🏛️💊 The White House wants to vet AI models like the FDA vets drugs.

The Move: Review what you actually finished last week — not what you worked on.

The Evidence: Research on implementation intentions shows that people systematically overestimate output while underestimating the friction of switching between tasks. A weekly done-list (not a to-do list) recalibrates that gap faster than any planning system.

Start Here: Spend 5 minutes right now listing every completed task from the past 7 days. Just what got done, nothing else. That list will tell you more about your real capacity than anything you planned.

Some people use Claude to write emails. Others use it to basically run their entire business while they play Wordle.

This isn't just ChatGPT's cooler cousin. It's the AI that's quietly revolutionizing how smart people work – writing entire business plans, planning marketing campaigns, and basically becoming the intern you never have to pay.

The Hustle's new guide shows you exactly how the AI-literate are leaving everyone else behind. Subscribe for instant access.

Get our Claude playbook

On May 5th, OpenAI launched a self-serve Ads Manager in beta — meaning any U.S. business can now sign up at ads.openai.com, set a budget, and run ads inside ChatGPT conversations. No agency required, no minimum spend. Six weeks ago, getting a ChatGPT ad required a $200,000 minimum commitment. That floor is now gone.

The mechanics are what you'd expect: cost-per-click and cost-per-impression bidding, a conversion pixel, and a Conversions API for tracking post-click actions. Agency partners include Omnicom, Publicis, WPP, and Dentsu. Tech partners include Adobe, Criteo, and StackAdapt. It's standard adtech infrastructure, assembled quickly.

What's interesting isn't the format — it's the context signal. When someone asks ChatGPT "which CRM should I buy for a 50-person sales team," they've disclosed budget, company size, and use case in a single exchange. That's a higher-intent signal than most search queries. OpenAI is betting that advertisers will pay a premium for it. Sam Altman once called advertising a "last resort." OpenAI needs $100 billion in annual ad revenue by 2030 to hit its projections. Last resort or not, the math made the decision.

Why it matters: ChatGPT is no longer just a product — it's a media channel. If you run paid acquisition, it's worth claiming an account this week, even if you don't spend yet.

Earlier this month, Anthropic launched Project Glasswing: a controlled initiative giving roughly 50 organizations — including AWS, Apple, Cisco, Google, JPMorgan Chase, Microsoft, and NVIDIA — access to Claude Mythos Preview, its most capable model to date. The reason it's restricted: in testing, Mythos autonomously identified and exploited zero-day vulnerabilities in every major operating system and every major web browser, including a 17-year-old remote code execution flaw in FreeBSD. No human direction after the initial prompt.

Anthropic isn't releasing Mythos publicly. The goal of Glasswing is to let defenders find and patch critical vulnerabilities before malicious actors develop comparable tools. Anthropic estimates similar capabilities will emerge from other labs within 6–18 months. The $100 million in usage credits committed to Glasswing partners suggests they're treating this window as genuinely short.

The more interesting structural note: Anthropic didn't explicitly train Mythos for cybersecurity. These capabilities emerged as a downstream consequence of general improvements in reasoning and coding. That's the part worth sitting with. The security implications aren't a feature — they're a byproduct of getting good at software.

Why it matters: We're entering a phase where AI capability upgrades arrive faster than the security infrastructure can respond. Glasswing is one answer. It probably won't be the last.

A quieter story, but the one with the longest tail: the architecture of software is shifting to assume AI agents as the main interface. Anthropic's Model Context Protocol, Salesforce's Headless 360 initiative, Zapier's programmable workflows, and new agent-focused infrastructure from Stripe and Mastercard all reflect the same thesis — that AI systems will increasingly be the entities making API calls, not humans typing into dashboards.

Industry analysts have started calling this "agent-first design." The argument is that websites, apps, and platforms will increasingly need to optimize not just for human users, but for autonomous AI systems navigating them. That has downstream effects on everything: SEO, UX, customer journeys, pricing pages, onboarding flows. A platform designed for a human who reads slowly and clicks through menus is a different thing than one designed for an agent that processes context and executes in parallel.

For creators and small operators, this is more of a heads-up than a call to action. The tools aren't fully there yet. But the direction is clear, and the infrastructure is being built now.

Why it matters: The way software gets used is changing at the architectural level. What worked for human-first design may not translate cleanly to an agent-first world.

The White House is drafting an executive order that would require AI models to be vetted before release — modeled, explicitly, on how the FDA approves drugs. National Economic Council Director Kevin Hassett cited Anthropic's Mythos model directly as the prompt. The Commerce Department has already expanded its voluntary testing program to include Google, Microsoft, xAI, OpenAI, and Anthropic.

The FDA comparison is doing a lot of work here. Drug approval can take a decade and cost hundreds of millions. AI development cycles are measured in months. Whether a regulatory framework designed for pharmaceutical timelines makes sense for a technology moving at this pace is, to put it charitably, an open question.

Why it matters: Regulation is no longer theoretical. The shape it takes over the next 12 months will define what AI labs can ship, and when.

→ The tools are reorganizing around AI. The Ultimate Productivity Playbook gives you a system for doing your best work inside the shift — not after it settles. Get it here →

📚 Read Next

The money moved this week. The platform ads launched, the model stayed locked, and the regulators took notice. That's three different industries acknowledging the same thing at the same time: this isn't a research project anymore.

Stay MOTIVATED,

Gio