A daily news analysis show on all things artificial intelligence. NLW looks at AI from multiple angles, from the explosion of creativity brought on by new tools like Midjourney and ChatGPT to the potential disruptions to work and industries as we know them to the great philosophical, ethical and practical questions of advanced general intelligence, alignment and x-risk.
Don't just listen to podcasts. Learn from them with transcripts, summaries, and chapters for every episode. Skim, search, and bookmark insights. Learn more
Claude Tag could mark a shift from AI as a separate app to AI as a persistent teammate inside the places teams already work. NLW breaks down five ways that could change how people use AI at work. In the headlines: Anthropic’s Fable fight, Meta model review, Chinese robots, Grok Build, and Seed Dance 2.5. Enterprise Agent Leadership Program (FKA EnterpriseClaw) - Next cohort begins 6.29.26: http://training.besuper.ai/ Brought to you by: KPMG – Research from KPMG and the University of Tex...
This episode addresses the growing controversy surrounding AI data centers, particularly concerns about their water and energy consumption. It debunks common misconceptions with factual data, highlighting that while local issues can arise, the overall impact is often exaggerated. The discussion advocates for a balanced approach where communities seriously engage with tech firms to negotiate substantial local benefits, moving past reductive debates to find win-win solutions.
This episode delves into the buzz around GLM 5.2, an open-weight AI model drawing comparisons to the DeepSeek R1 moment due to its surprising real-world performance, especially in coding and web design. It explores how this challenges the perceived duopoly of OpenAI and Anthropic, forcing enterprises to rethink their AI strategies beyond state-of-the-art models. The discussion also touches on the Fable V ban, Trump's views on AI regulation, and a significant talent exodus from Google DeepMind.
Nathaniel Whittemore and Nufar Gaspar discuss why companies should rethink their dependence on frontier cloud models, citing rising token costs, vendor fragility, and data control issues. They provide a practical primer on local AI, breaking down its basic layers from hardware and open models to serving tools like Ollama and agent harnesses. The episode highlights the benefits and trade-offs of running AI on controlled machines, offering guidance for executives, practitioners, and enthusiasts alike.
This week marked a significant realignment in the AI industry, primarily triggered by Anthropics suspending Fable V access due to US export controls. This incident underscored the risks of building strategies around a single frontier model, leading to increased attention on open-source alternatives like GLM 5.2 and innovative model routing architectures such as OpenRouter's Fusion API. The shift reflects a growing demand for local control and a more fragmented, strategic, and contested AI ecosystem. Even SpaceX's acquisition of Cursor and Europe's AI sovereignty scramble point to this broader industry transformation.
The episode argues that the recent Fable V disruption highlights the need for companies to move beyond vendor-centric AI strategies to build robust AI learning systems. It covers the White House and Anthropic's ongoing discussions on AI security flaws, Bernie Sanders' controversial proposal for a $7 trillion AI sovereign wealth fund, and Accenture's recent financial challenges due to AI transformation. The core discussion centers on Satya Nadella's concept of "human capital" and "token capital," emphasizing the importance of institutional learning loops and model-portable IP for a stable AI ecosystem.
Following the Fable V shutdown, the G7 debated AI access and international cooperation, highlighting European concerns over US export controls and their impact on tech sovereignty. The AI industry is now focused on finding alternatives, with Chinese open models gaining traction and enterprises exploring multi-model architectures and smart routing strategies to achieve high performance while managing costs and ensuring predictable access. This shift accelerates the move towards model diversity and smarter enterprise AI.
This episode explores a significant shift in the AI race, beginning with the ongoing regulatory dispute between Anthropic and Washington concerning the Fable V/Mythos model, highlighting communication breakdowns and the implications of an ad-hoc regulatory regime. It then pivots to Elon Musk's strategic moves, detailing how SpaceX's IPO success and "neo-cloud" revenue are fueling XAI's expansion through the $60 billion Cursor acquisition, and the ambition for new, powerful models. Finally, the episode delves into OpenAI's recently leaked financials, clarifying that significant reported losses are largely due to non-cash accounting changes, and revealing strong inference profits, which may influence their IPO timeline.
This episode argues that mass-scale AI training is crucial for sustaining the American economy, which is increasingly driven by AI infrastructure investment. It explores the shift from basic, seat-based AI usage to agentic, usage-based consumption, driving immense revenue for AI labs but leading to enterprise cost scrutiny and token scarcity. The solution lies in equipping workers with the skills to effectively use and manage AI agents, moving beyond simple productivity to unlock new economic value and meet the growth demands of AI labs.
The podcast details the ongoing Fable V crisis, where Anthropic's AI model was shut down after Amazon reported a jailbreak to the US government, citing national security concerns. Conflicting narratives have emerged regarding the severity of the jailbreak and Anthropic's response, with some experts arguing the issue was political rather than technical. The episode explores the intense disagreements between Anthropic and the White House, the role of key decision-makers, and the broader implications for AI policy and innovation in the US.
This week's AI brief highlights the dramatic events surrounding Anthropic's Fable V, which was first hailed as a powerful new model and then embroiled in controversy over foreign national access and restrictive policies. The discussion delves into its advanced capabilities and the uproar over undisclosed nerfs for LLM research, underscoring the significant power frontier AI labs wield. Additionally, the episode touches on the successful SpaceX IPO and the growing concern of "token panic" as companies begin to cap employee AI consumption.
This episode details the stunning news that the US government ordered Anthropic to suspend access to its Fable V and Mythos V AI models for foreign nationals, forcing a global shutdown. The host explores Anthropic's defense against the alleged jailbreak, the widespread backlash from the AI community against both the government's overreaction and Anthropic's past "safetyism," and the profound implications for future AI development, market stability, and international technological sovereignty. This incident marks a major turning point for government control over frontier AI.
This episode debunks the panic surrounding a viral 'AI bubble' chart, arguing it reflects market rationalization and the shift to token scarcity rather than collapsing demand. The host analyzes the chart's true meaning as a price index, highlighting the vast growth potential in AI adoption, especially among less mature users. Key industry headlines, including SpaceX's IPO, Bezos' Prometheus AI, Meta's Manus split, chip supply chain issues, and Goldman Sachs' bold AI infrastructure forecast, are also covered, all pointing to an expanding AI market.
The Fable 5 AI model release by Anthropic has become the most controversial to date, facing backlash for its strict safety guardrails, extended data retention policy, and especially the 'silent nerfing' of AI development capabilities. This incident ignited a significant debate about the extent to which leading AI labs should dictate how users can build, study, or access AI technologies. The episode also covers headlines on Trump's AI equity proposal, OpenAI's massive Ohio data center plans and growing public resistance to data center construction, and Broadcom's new AI compute financing fund.
This episode delves into the launch of Anthropic's Fable 5, hailed as the best AI model to date, showcasing its groundbreaking benchmark performance and real-world transformative capabilities. It explores user experiences ranging from complex coding projects to strategic ideation, highlighting a shift from giving AI tasks to delegating responsibilities. The discussion also addresses controversies surrounding strict guardrails, limitations on AI research, and data retention policies, alongside new skills needed for optimal AI utilization.
This episode delves into OpenAI's declaration of a new AI phase centered on automated research, accelerating the economy, and providing personal AGI. It also covers major headlines including OpenAI and SpaceX's IPO filings, Intel's emerging role in AI chip manufacturing amid supply shortages, and Washington's intensifying AI regulation debate. A key discussion point explores whether "AI" is bifurcating into distinct consumer AI and work AI categories, questioning the industry's future direction.
ChatGPT’s rumored “super app” overhaul isn’t just an IPO story — it’s a sign that AI use is shifting from chat to agents, coding tools, and loops. The result is a widening advantage gap between casual users and power users, with agent users seeing compounding gains while regular chat users stay linear. In the headlines: Trump explores government stakes in AI labs, Google rents SpaceX compute, and NVIDIA secures SK Hynix memory supply. Brought to you by: KPMG – Research from KPMG and the Universi...
AI is making it possible to build richer versions of the files knowledge workers send every day: decks, memos, spreadsheets, reports, proposals, training materials, and more. This has gotten even easier this week with the release of OpenAI's "Sites" feature in Codex. In this practical Operator's episode, NLW walks through 10+ examples of work outputs that are often better as living, shareable, updateable, interactive links than static documents. Sign up for AI Executive Catchup: https://aiexe...
The AI landscape is rapidly shifting into a token-efficiency era, with companies adapting to usage-based models and developing solutions to cut costs while maintaining performance. Noteworthy innovations include Codex's expanded plugin ecosystem and the 'Sites' feature, enabling one-click web app creation from knowledge work. Furthermore, the debate around AI ownership is escalating, with policymakers considering government equity stakes in major AI labs. Enterprises are urged to implement architectural changes and robust training programs to manage token costs effectively.
Nathaniel Whittemore breaks down significant developments in the AI world, covering the U.S. government's discussions on taking equity stakes in AI labs, OpenAI's major "Dreaming" memory system update for ChatGPT, and industry challenges like TSMC's chip shortage. The episode also delves into Anthropic's "When AI Builds Itself" paper, exploring recursive self-improvement and future scenarios, alongside critical reactions. Finally, it examines OpenAI's proposals for frontier AI governance and bipartisan efforts in Congress to regulate the technology.
This episode explores significant shifts in the AI landscape, starting with ChatGPT's rapid ascent to a billion users and the growing dominance of AI bots in web traffic. It then examines Meta's strategic move into business AI with new agents aimed at small enterprises. The core discussion revolves around token efficiency, highlighting its emergence as the primary metric for enterprise AI success over raw intelligence, driven by rising costs and supply constraints. The episode concludes by detailing how AI labs, product companies, and architectural innovations are addressing these efficiency challenges through smart routing and continual learning to optimize AI implementation.
The episode delves into the evolving landscape of enterprise AI, highlighting OpenAI's push to integrate Codex into broader knowledge work with new features like annotations, role-specific plugins, and "Sites." Simultaneously, Microsoft's Build announcements emphasize cost-optimized, customizable frontier models as a solution to the growing "token scarcity." The discussion also touches upon a contentious AI executive order and the industry's response to the global memory chip shortage.
The episode delves into the burgeoning AI market, from Anthropic and OpenAI's anticipated IPOs to Google's massive equity raise for AI buildout. It also highlights challenges like the Instagram AI support exploit and companies struggling with AI ROI. A key focus is the evolving policy debate, including Bernie Sanders' proposal for public ownership in AI and broader discussions on distributing AI's financial upside as a public good.
May 2026 marked a critical turning point in AI, transitioning from an era of subsidized usage to one defined by token scarcity and rising costs. This shift has led to "AI sticker shock" for enterprises, prompting a move towards usage-based billing from major model providers like GitHub Copilot, Google, and Anthropic. The episode delves into how companies are adapting to these new economic realities, from managing costs and seeking external support to strategic infrastructure realignments, including Elon Musk's new role in providing compute and the broader market's response to the token shortage.
A practical primer on /goal, the new AI primitive showing up in Codex and Claude Code. NLW explains how /goal differs from a normal prompt, why it matters for longer-running agent tasks, what makes a good goal, and how to think about using it beyond coding for audits, research, vendor reviews, market landscapes, and other knowledge work where the AI needs a clear finish line and evidence of completion. Sign up for AI Executive Catchup: https://aiexecutivecatchup.com/ Brought to you by: KPMG – ...
The AI Daily Brief dives into Anthropic's Claude Opus 4.8, a modest yet meaningful upgrade featuring better judgment, stronger self-checking, and reduced bluffing. The episode compares Opus 4.8 to GPT-5.5, discusses the new dynamic workflows in Claude Code, and emphasizes the growing importance of the model harness. Additionally, it covers significant AI industry headlines, including a mega law firm's $500M internal AI investment, Cognition's $26B valuation, Meta's AI cloud ambitions, and Microsoft's upcoming model releases.
Nathaniel Whittemore delves into the increasing discussion around an AI token tax, analyzing various proposals from figures like Elizabeth Warren, Mark Cuban, and Dario Amodei, who suggest taxing AI to fund public services or support displaced workers. The episode presents the foundational economic argument for shifting the tax base as productive capacity moves from humans to AI agents. However, it also thoroughly critiques the token tax, highlighting concerns about tokens as an unreliable value proxy, potential negative impacts on AI experimentation, and the risk of entrenching large incumbents, advocating for a nuanced and open debate.
This episode addresses the annual "AI slowdown panic," exploring how token shortages, usage-based pricing, and agent cost overruns are shaping the market. It discusses a new, more robust coding benchmark, evolving perspectives on AI's impact on job displacement, and significant funding for the AI inference layer. The conversation ultimately argues that current constraints reflect a market learning to price scarce compute rather than collapsing demand.
Pope Leo XIV's encyclical, "Magnifica Humanitas," offers a nuanced Catholic perspective on artificial intelligence, asserting that AI is not inherently evil but never morally neutral, and stressing that human value transcends mere intelligence or productivity. The document champions human dignity, warns against potential data exploitation and a new form of colonialism, and fundamentally argues for AI's categorical difference from human beings, aiming to plant a flag for future debates on AI's societal role. The episode also covers key AI industry updates including Anthropic's Mythos, intelligence agency AI initiatives, DeepSeek's pricing, and Grok's latest model.
Nufar Gaspar joins NLW to explain how executive AI usage significantly predicts broader organizational adoption. They reveal common pitfalls leaders face and introduce five non-negotiable operating principles for effective AI interaction. The discussion then details four key AI team members executives can implement today, providing practical tips for each: a research analyst for informed decisions, a strategic thought partner for unbiased guidance, a communication expert to refine voice, and an operational powerhouse for enhanced visibility and support.
