A weekend scramble over Anthropic’s newly released Claude Fable 5 became a test case for how the Trump administration may regulate AI: through formal safety policy or through panicked, personalized power. Nilay Patel and Verge reporter Hayden Field trace the model’s abrupt shutdown, the export controls that followed, and the industry’s growing fear that political risk now sits alongside technical risk.

Rob Ayers traces how deployment practices evolved from weekly release boards to progressive delivery, arguing that feature toggles, roll-forwards, and platform teams do more to reduce risk than ritualized rollbacks. The conversation also untangles GitOps beyond its name, surveys Kubernetes at enterprise scale, and considers how AI-driven code velocity may shift CI/CD from speed toward safety.

GitHub’s Kyle Daigle describes a software world remade by coding agents, where pull requests and commits are surging, nondevelopers are building apps, and maintainers need new controls to keep pace. The conversation follows the business and product consequences of that shift, from model routing and token costs to the long game of training tools on a developer’s habits and context.

As AI coding tools shift from one-off prompts to self-directed automations, the conversation breaks down loops in plain English: scheduled tasks, hooks and goal-based agents that keep working until a job is done. Working through Claude Code and Codex, it also makes the case for guardrails, isolated workspaces and clear validation before handing autonomous agents the keys.

At Platformer’s first live show, Casey Newton and guests take stock of AI’s contradictory role in the workplace: a source of anxiety, a generator of strange new job titles, and a tool that may make software creation radically more personal. The conversation moves from Atlassian’s enterprise “context graph” to Eugenia Kuyda’s vision of companions and bespoke apps, tracing how automation may reshape both office work and the products people use every day.

Claire Vo and Braintrust CEO Ankur Goyal make the case that coding agents belong in the hardest engineering work, from query optimization and data migrations to the unglamorous discipline of CI. Their conversation argues that the real leverage comes from rigorous evals: defining success clearly, letting models explore the how, and using systems to extend human judgment rather than replace it.

Skydio chief executive Adam Bry argues that autonomous drones are moving from camera toys to networked infrastructure, with their most consequential uses in emergency response, utility inspection, and military reconnaissance. The conversation turns on whether the United States can build a domestic drone industry at scale while navigating bans on Chinese rivals, the politics of surveillance, and the moral limits of AI in warfare.

A sudden U.S.-ordered shutdown of Anthropic’s Fable 5 has set off a fight over AI safety, export controls and who gets to decide when a jailbreak justifies pulling a frontier model offline. The conversation weighs incompetence against political theater, while warning that if the ban sticks, access to advanced AI could hinge on nationality, ID checks and a far broader crackdown across the industry.

Financial Times writer Sarah O’Connor argues that AI’s effect on work is not an unstoppable natural force but a series of choices shaped by power, policy and the people closest to the job. Drawing on reporting from translators, nurses, software developers and Hollywood writers, she makes the case that the real fight is over human agency and what kinds of work should remain irreducibly human.

Condé Nast CEO Roger Lynch argues that legacy media survives by acting less like a magazine publisher than a portfolio of brands built for events, subscriptions, commerce, video, and adaptation. He talks through the company’s restructuring, the collapse of Google-driven traffic, the uneasy bargains of AI licensing, and why authority matters more than raw scale in the creator era.

Mike Krieger describes how living with Anthropic’s new Fable 5 model reshapes work from prompting to delegation, turning AI from a clever assistant into something closer to a teammate. The conversation follows what that shift means for software engineering, verification, workplace collaboration and the widening gap between a person’s intent and their ability to build.

Brookings fellow Molly Kinder argues that the real danger of AI is not an instant jobs apocalypse but a long, destabilizing period of selective white-collar displacement for which government and industry have no credible plan. The conversation weighs retraining, safety nets, and broader wealth-sharing proposals against a future in which automation chips away at the skills and careers workers once thought were secure.

Mustafa Suleiman argues that Microsoft’s evolving partnership with OpenAI has pushed the company toward model self-sufficiency, even as he insists the alliance remains central to its AI strategy. In a wide-ranging conversation, he defends the coming wave of enterprise automation, rejects claims of machine consciousness, and says the technology will have to prove itself by making people healthier, happier, and more capable.

Tony Fadell argues that breakthrough products come from identifying real pain, pairing it with newly viable technology, and shaping the entire customer journey, from interface to marketing story. He warns that AI can accelerate prototyping but cannot replace human judgment, taste, and responsibility without leaving builders with brittle products and long-term debt.

Nilay Patel and New York Times reporter Ryan Mac sift through SpaceX’s blockbuster IPO filing to ask what Elon Musk’s ownership of X has actually become and why the company’s shrinking social platform may not matter. Their conversation turns into a broader indictment of a market structure that keeps rewarding Musk with more power even as corporate governance, shareholder accountability, and business fundamentals erode.

A conversation about the so-called SaaS apocalypse argues that AI will not kill software so much as multiply it, pushing more people into building while preserving the value of products that handle maintenance, context and trust. Along the way, the speakers compare homemade agents, email triage, design systems and Figma’s bet that the future belongs to proactive, personalized tools that move fluidly between code and design.

Claire Ho tries to turn herself into an AI-generated on-screen character using Google Flow and Gemini’s video tools, building a glossy podcast hype reel while narrating every glitch, typo, and uncanny surprise. The experiment doubles as a case study in how multimodal AI can act less like a coding assistant than a rough-cut creative producer.

Alex Kantrowitz and MG Siegler parse an uneasy moment for Big Tech, from Google’s lagging AI product strategy to Apple’s foldable ambitions and Meta’s muddled subscription push. The conversation argues that agents and chatbots are converging into a new interface for the web, one that could reorder who controls computing itself.

Recording Academy chief Harvey Mason Jr. sketches an industry already saturated with AI, where songwriters use models for everything from chord progressions to demo vocals even as the Grammys try to preserve a meaningful line around human authorship. The conversation also turns to the academy’s move from CBS to Disney, the politics of platform power, and the scramble to keep music culture visible in the age of TikTok.

Benedict Evans argues that AI is a platform shift on the scale of the internet or mobile: transformative, messy, and still too early for anyone to know where the real value or disruption will settle. He pushes back on jobpocalypse panic, sketching a slower, more uneven reshaping of work in which adoption, distribution, and new kinds of services matter more than apocalyptic forecasts.