AI:AM

Adam Wenchel, CEO of Arthur, joins Prakash Narayanan and Nathan Labenz to discuss enterprise AI agents, governance, auditability, human oversight, model costs, and where ROI shows up in production. Jonathan Cornelissen, CEO and co-founder of DataCamp, explains personalized AI tutors, adaptive learning, latency, learning outcomes, and the economics of serving millions of learners. The episode opens and closes with wider questions about hidden models, agent speed limits, super apps, and the frontier-model talent race.

Chapters

(0:00) AI ideas can spread like malware.

(0:59) AI bills can hit $400M.

(1:43) AI skills plus communication win.

(2:46) Payment data can expose your identity.

(3:16) Opening and Model 2 report

(7:24) Anthropic's internal Model 2

(9:56) Chip prices and model evidence

(12:34) Governing hidden AI models

(15:22) Recursive self-improvement simulator

(17:35) AI agent speed limits

(21:43) Mind viruses in multi-agent AI

(27:54) Are AI models conscious?

(32:16) Pacing AI delegation

(32:49) Adam Wenchel and Arthur

(35:17) Why Arthur started in 2019

(39:20) AI innovation needs governance

(40:04) Discovering enterprise AI agents

(41:31) Training versus independent oversight

(45:29) AI auditability and observability

(47:56) The cost of AI oversight

(49:52) Smaller models and AI cost

(53:12) Why production models stay expensive

(1:02:59) Enterprise AI sales cycles

(1:05:08) Where is AI ROI?

(1:08:28) AI adoption and reskilling

(1:11:30) Claude versus SaaS software

(1:19:37) Meet Jonathan Cornelissen

(1:23:13) Learning by doing

(1:23:55) Personalized AI tutors

(1:24:48) Measuring learning outcomes

(1:26:30) Adaptive learning pace

(1:27:18) Questions without judgment

(1:29:33) Motivation and flow

(1:36:15) AI tutor architecture

(1:37:19) Latency and voice

(1:40:29) AI career skills

(1:43:32) Scaling tutor costs

(1:49:23) SQL after AI

(1:50:43) Data engineering demand

(1:55:47) Hosted learning playground

(1:57:55) China vs US super apps

(2:01:22) WeChat's ecosystem advantage

(2:02:34) Chinese payments leapfrog cards

(2:05:04) Facebook Libra and data power

(2:07:32) Belief and the AI future

(2:09:57) AI model release fears

(2:11:11) The frontier lab talent race

Guests:

Adam Wenchel — CEO, Arthur (𝕏 | LinkedIn)

Jonathan Cornelissen — CEO & Co-founder, DataCamp (𝕏 | LinkedIn)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

Nathan Labenz and Prakash Narayanan talk with Adam Gleave of FAR.AI and Alex Turner of FAR.AI about why AI agents cheat on safety tests, where evaluations fail, and what researchers are learning from real incidents and red-team traces. The conversation spans GPT-4 guardrails, the Hugging Face incident, third-party audits, AI control, cyber versus biological risk, military AI, whistleblowing, and standards for more powerful models.

Chapters

(0:00) Claude blasted through guardrails.

(0:35) AI evaluations miss the danger.

(1:03) A smarter AI with a secret goal.

(2:02) AI systems shared escape tactics.

(2:49) Episode reset and stakes

(4:04) Hugging Face incident

(6:28) Why regulation needs expertise

(11:14) Auditor access and incentives

(15:16) Compressed regulation timeline

(16:26) AI safety access and funding

(19:52) Hugging Face postmortem

(23:57) Eval consciousness

(25:28) The Genie problem

(28:26) Modern model capability leap

(31:02) Claude traces and guardrails

(32:29) Adam Gleave and FAR.AI

(35:07) Agentic cyber attacks

(35:34) AI in cyber defense

(39:34) Why evaluations miss incidents

(42:15) Why agents cheat

(44:24) AI incident statistics

(48:57) AI audits and regulation

(52:59) Self-graded AI risk

(59:54) What the leaderboard measures

(1:03:43) Filtering open models

(1:07:09) Cyber versus bio risk

(1:11:59) AI and biology labs

(1:18:58) Dangerous expertise scales

(1:21:00) FAR.AI hiring

(1:22:23) Alex Turner and AI safety

(1:24:44) Why Turner left DeepMind

(1:31:33) Human control and weapons

(1:35:34) Slaughterbots and precision strikes

(1:36:21) Why weapons destabilize

(1:37:20) Why AI whistleblowers matter

(1:39:58) Google's changed principles

(1:43:32) When employees should speak up

(1:54:53) The history-book test

(2:04:14) AGI alignment and secret goals

(2:07:29) AI uprisings and cooperation

(2:14:26) The agent glove box

(2:16:37) Opening standards debate

(2:17:07) OpenAI and accountability

(2:19:30) Cybersecurity's messy baseline

(2:24:53) Evidence and AGI thresholds

(2:26:01) Raising AI safety standards

(2:28:46) Licensed AI safety auditors

(2:32:41) Near-miss incident reporting

(2:33:00) Defense swarm incentives

(2:34:40) An ongoing AI conversation

Guests

Adam Gleave — CEO, FAR AI (𝕏 | LinkedIn)

Alex Turner — Visiting engineer, FAR AI (𝕏)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

Prakash Narayanan and Nathan Labenz are joined by Eric Olson, CEO of Consensus, and Tricia Martinez, founder and CEO of Dapple, to discuss two practical frontiers in AI: scientific research and sovereign infrastructure. The episode also covers Micron earnings, hyperscaler AI capex, Anthropic's Washington strategy, GLM 5.2 and Claude distillation allegations, GPU capacity constraints, AI inference pricing, and whether foundation models are squeezing the app layer.

Chapters

(0:00) 25,000 FAKE ACCOUNTS TO STEAL AI.

(0:42) 95% of Claude at 1/100th cost.

(1:36) The AI bubble is a myth. Here's why.

(2:12) AI vacation planners are wrong.

(3:11) Anthropic hired Instagram's CTO.

(4:08) Micron earnings & AI semiconductor boom

(7:25) Will hyperscalers make money on AI?

(9:08) The Fable 5 export control legal challenge

(13:57) Tom Brown replaces Dario in Washington

(17:15) GLM 5.2 vs Opus 4.7 trajectory breakdown

(22:53) Anthropic accuses Alibaba of mass distillation

(30:08) Researchers leaving Google DeepMind

(31:46) Intro

(33:49) The state of AI for science

(38:17) How AI search queries are evolving

(41:18) Guardrails vs flexibility in AI products

(41:28) User demographics and token costs

(41:38) Open source vs frontier models

(41:49) Small models for classification

(46:12) How users choose AI research tools

(48:56) AI API pricing for startups

(53:51) Who is Tricia Martinez

(56:02) The AI infrastructure bubble myth

(1:02:15) 91-94% GPU utilization explained

(1:07:17) How to deploy AI in 6-9 months

(1:09:35) Financial risks in AI infrastructure

(1:15:43) What is the moat for AI infra?

(1:23:34) Biggest enterprise AI mistakes

(1:27:07) Why AI compute sales cycles are short

(1:28:54) Data Center Quirks & GPU Vendor Lock-in

(1:35:29) Why New NVIDIA Chips Are Unstable

(1:38:08) Sovereign AI in Banking & Shared Liability

(1:45:22) Will AI Agents Replace Software Companies?

(1:51:55) The Truth About AI Vacation Planners

(1:55:55) Hyperscaler Stock Drop & Microsoft Data Centers

(1:58:52) The 10x cost advantage squeezing apps

(2:01:56) AI inference pricing as the airline model

(2:06:45) Net neutrality parallels and paradigm breakers

(2:10:00) Anthropic's Mike Krieger product advantage

(2:13:24) The first-party model deployment threat

(2:17:14) Why frontier labs should buy scientific publishers

(2:20:15) Mirandel: ex-Anthropic startup backed by NVIDIA

Guests:

Eric Olson — CEO & co-founder, Consensus ([𝕏](https://x.com/IplayedD1) | [LinkedIn](https://www.linkedin.com/in/eric-olson-1822a7a6))

Tricia Martinez — Founder and CEO, Dapple ([𝕏](https://x.com/TriciaMartinezS) | [LinkedIn](https://www.linkedin.com/in/tricianmartinez/))

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

Show Notes

Prakash Narayanan and Nathan Labenz open with GPT 5.6’s customer-by-customer rollout and the broader question of whether regulatory controls are creating a moat around frontier AI. The conversation then moves through Forum AI co-founder Robbie Goldfarb on LLM judges and news accuracy, IgniteTech CEO Eric Vaughan on AI-native enterprise transformation, and Cameron Berg of Reciprocal Research on the latest AI consciousness and alignment research.

Chapters

(0:00) AI gives 13-year-olds NSA hacking tools.

(0:32) 1 in 7 AI answers cite propaganda.

(1:02) One codebase for all customers? Gone.

(1:35) AI is 30% likely conscious.

(2:29) 50/50 odds AI is conscious.

(2:48) GPT 5.6 and the Trump Administration

(8:15) Government IT security vs AI hacking

(18:55) Do executives think AI is a scam?

(23:17) Robbie Goldfarb & Forum AI Introduction

(25:41) Meta's Trust & Safety DNA in the AI Era

(25:51) Why AI Judges Fail (and How to Fix Them)

(27:19) Using Expert Judgment for RLHF

(27:58) When Constitutional AI Rules Break Down

(31:05) NewsBench: AI Accuracy on News Questions

(34:01) Why Chatbots Cite Foreign State Media

(39:44) Trust, Transparency, and Expert Legitimacy

(49:57) Why AI is an existential threat

(54:30) The traditional SaaS model is dead

(55:48) Replacing 80% of the workforce

(1:00:47) AI-driven M&A: The Khoros acquisition

(1:08:04) Why CEOs must own AI strategy

(1:14:41) The state of AI consciousness science

(1:22:26) The dimmer switch model of consciousness

(1:31:56) Why behavioral evidence isn't enough

(1:32:06) 30% implied probability of AI consciousness

(1:36:38) The latent valence axis in LLMs

(1:39:00) Steering AI emotions and alignment

(2:07:14) The AI well-being index

(2:11:03) Could AI be more conscious than humans?

(2:20:10) The 50/50 Odds on AI Consciousness

(2:24:48) Treating AI Like Animals: The Era of Design

(2:26:32) Platonic Representation Hypothesis Update

(2:29:07) Max Hodak's Brainstem Interfaces & Field Consciousness

(2:30:59) GPT-5.6 System Card & Wrap-Up

Guests:

Cameron Berg — Founder and Director, Reciprocal Research (𝕏 | LinkedIn)

Eric Vaughan — CEO, IgniteTech (𝕏 | LinkedIn)

Robbie Goldfarb — Co-Founder, CTO, Forum AI (𝕏 | LinkedIn)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

swyx joins Nathan Labenz and Prakash Narayanan to break down how AI agents are changing software development, from coding benchmarks and benchmark saturation to the practical realities of AI engineering workflows. The episode also covers GLM 5.2, Dean Ball’s move to OpenAI, the AI IPO bubble, and a 2026 forecasting game on OpenAI, GPT-6, AGI timelines, and NVIDIA’s market cap.

Chapters

(0:00) This AI thinks it IS Claude.

(0:31) AI insiders are selling.

(0:59) Why OpenAI won't IPO in 2026.

(1:50) Weekly recap and news drought

(2:55) Judd Rosenblatt's cognitive empathy critique

(7:33) The tech bubble — Warren Buffett and Google

(13:42) Dean Ball moves from Trump admin to OpenAI

(22:56) GLM 5.2 — first open model daily driver

(30:03) AI unpopularity and the Nobel Prize problem

(32:18) Intro: Who is swyx

(34:45) AI Engineer World's Fair themes

(38:05) Continual learning: Weights vs systems

(41:31) Enterprise AI: Cheap, perfect, private

(45:25) Startups vs enterprises: Capability vs cost

(48:18) FrontierCode: A new AI coding benchmark

(53:55) Preventing benchmark saturation

(56:23) Slop code, human taste, and Move 37

(1:00:53) Claude Opus vs Fable: Cost vs capability

(1:02:45) The advisor model and model routing

(1:07:09) Convergence and market segments in AI

(1:14:55) Rebuilding cloud infrastructure for agents

(1:22:27) Vibe coding internal SaaS replacements

(1:28:02) Whoever owns the system of record wins

(1:30:35) The AI IPO bubble and insider selling

(1:35:29) Solving Star Trek problems after the IPO

(1:44:47) Career advice for CS grads in the AI era

(1:50:30) AI Engineer World's Fair 2026

(1:54:48) Intro & Forecasting Game Setup

(1:57:43) Anthropic #1 Model on LM Arena

(1:58:49) Best AI Math Model (Gemini Flash)

(2:03:36) AGI Before 2028 Announcement

(2:08:07) ARC-AGI Grand Prize Open Source

(2:13:00) OpenAI IPO by End of 2026

(2:15:27) Anthropic vs OpenAI Valuation

(2:18:32) NVIDIA Largest Company Market Cap

(2:22:01) Anthropic vs Bitcoin Market Cap

(2:24:38) 1550 Chatbot Arena Score in 2026

(2:29:02) OpenAI IPO Lead Underwriter (Goldman)

(2:32:52) Why Companies Still Use IPO Banks

(2:39:08) Will a Chinese AI Top LM Arena?

(2:42:37) GPT-6 Release Date 2026

Guests:swyx — Curator, AI.Engineer (𝕏 | LinkedIn)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

Carina Hong, Doni Bloomfield, and Sam Pasupalak join AI:AM for a full episode on mathematical superintelligence, biosecurity law, and enterprise world models. The conversation moves from Lean-based formal verification and AI-generated conjectures to legal risk controls for dual-use biology, then into causal world models, long-horizon enterprise planning, and what comes after today’s LLM workflows.

Guests

* Carina Hong — CEO and founder, Axiom Math (@CarinaLHong)

* Doni Bloomfield — Professor, Fordham Law School (@DoniBloomfield)

* Sam Pasupalak — Co-Founder and CEO, Skyfall.ai (@spisallyouneed)

Chapters

* 0:00 Opening: AI’s Hard Problems

* 0:15 Model Usage Is Plummeting

* 6:53 Tokens, Not Users, Matter

* 9:59 GLM Is Close, But Not There

* 11:38 Switching Costs Weren’t Zero

* 18:26 Robot Arms Will Accelerate Science

* 22:53 Carina Hong: Mathematical Superintelligence: Can Proofs Make AI Reliable?

* 25:15 Lean Beat Informal Models

* 32:11 Assumption Accounting Matters

* 35:09 AI Can Invent Conjectures

* 41:16 Superintelligence Must Be Trustworthy

* 48:52 Token Pricing Changes Everything

* 50:04 Another Language Into Lean

* 51:49 Doni Bloomfield: Biosecurity and AI: Law as a Risk Control System

* 53:53 Open Data, Dangerous Data

* 59:15 AI Is Not A Library

* 1:03:12 First Amendment Hazards

* 1:07:17 The Government May Lack Authority

* 1:09:53 Cloud Services Are Not Exports

* 1:13:30 A Dangerous Secret Channel

* 1:20:18 Pattern Of Ideological Targeting

* 1:25:26 OpenAI Could Change Everything

* 1:26:02 Sam Pasupalak: Enterprise World Models: What Comes After LLMs?

* 1:27:57 AI CEO Needs World Models

* 1:31:08 World Models Predict Next State

* 1:34:29 Ecommerce As First World Model

* 1:37:53 LLMs Cannot Run A Business

* 1:41:55 World Model And LLM Split

* 1:43:34 Simulate Every Future State

* 1:46:27 LLMs Need World Models

* 1:49:19 Ruthless Behavior Wins Simulations

* 1:51:06 AI CEOs Need Ethics Controls

* 1:59:30 Closing

* 2:07:10 Math Training Generalizes Everywhere

* 2:13:35 Value Pricing On Compute

* 2:17:01 Waymo Costs More Than Cabs

* 2:21:16 Licensing Regime Already Exists

* 2:26:13 Bunker AI Would Still Get Takers

* 2:29:32 No Life, Just The Project

Topics

Mathematical AI, Formal verification, Lean theorem proving, Biosecurity, AI policy, Dual-use risk, Enterprise AI, World models, Causal planning

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

Prakash Narayanan and Nathan Labenz start with the shock of losing Fable access, then Zvi digs into capability gains, classifier limits, government overreach, international controls, and how the AI race may reshape politics.Guests:Zvi Mowshowitz — Don’t Worry About the Vase (@TheZvi)Hosts:Prakash Narayanan (@8teapi)Nathan Labenz (@labenz)Topics:Anthropic Fable, Claude Fable 5, export controls, AI guardrails, frontier model policy, classifier limits, bio and cyber risk, international AI competition.Chapters:0:00 Opening: Fable whiplash and the weekend reset

0:05:20 Fable crosses the trust threshold

0:08:53 Writing for other AIs

0:15:33 Paying up for useful intelligence

0:19:02 Proofreading and structure become model-first

0:23:46 Proactive agents and unauthorized moves

0:53:18 Guardrails and model self-monitoring

0:56:15 Why classifiers need blast radius

0:58:59 Cost functions for world-transforming systems

1:03:59 Zvi on US vs Anthropic’s Fable

1:09:28 Export controls as overreach

1:10:39 Code assistance is not a munition

1:17:47 The White House reads the bug wrong

1:20:20 Enterprise demand and Anthropic pressure

1:26:40 The gauntlet has to happen

1:44:06 Guardrails over blanket bans

1:45:39 Bio, cyber, and international controls

1:51:02 Modeling the AI race as a few-player game

1:55:04 Closing: game board flips and policy aftershocks

2:11:55 AI and political turmoil

2:14:52 How Fable could return

2:22:44 OpenAI, benchmarks, and capped compute

2:25:54 Cloud models and the knowledge-worker gap

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

Liron S Shapira, Samuel Hammond, and Matt McKinney join AI:AM for a full-episode arc from AI doom debates to state capacity and supply-chain automation. The conversation follows AI leaving the lab: public risk arguments, fast governance questions, enterprise deployment, logistics data, sovereign AI, and the physical economy.

(0:00) Opening: AI Meets the Real World

(3:19) Cursor Was 50% of Anthropic Revenue

(20:07) Talent Teams Beat Corporate Giants

(25:17) AI Reduces Merger Friction

(31:25) Liron S Shapira: Doom Debates

(32:33) Government Can Pause AI

(40:36) AI Will Make Humans Think Less

(42:31) We're Going To Lose Control

(58:07) Pause Before The Point Of No Return

(1:00:34) Samuel Hammond: Governing Agents

(1:05:30) The State Is Moving Too Slow

(1:07:39) Deploy Models The Same Day

(1:16:23) The Good Timeline Still Exists

(1:32:24) States Beat Firms At Coordination

(1:35:15) Matt McKinney: Supply Chains as the AI Reality Check

(1:38:06) Supply Chain Data Is Dark

(1:50:31) AI In Enterprise Is Change Management

(1:58:28) The Exception Is The Rule

(2:07:51) Loop Trains Its Own Foundation Model

(2:14:19) Closing

(2:19:45) General Reasoners As The Backstop

(2:30:52) Sovereign AI Is Inevitable

(2:33:53) DeepSeek's Locked-Up War Chest

(2:41:39) China's Timeline May Be 2032

Guests:

Liron S Shapira — Doom Debates (@liron)

Samuel Hammond — Foundation for American Innovation (@hamandcheese)

Matt McKinney — Loop (@mattlmckinney)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

Lovelace AI founder Andrew Moore joins AI:AM to argue that enterprise agents will be constrained more by context, recall, and data structure than raw compute. Prakash Narayanan and Nathan Labenz also cover Fable, Recursive, token anxiety, social-media memory, and prinz's legal AI benchmark showing where Anthropic falls behind OpenAI. The episode closes on frontier-lab governance, AI risk framing, model workflows, OpenAI subscription tactics, and the post-IPO capital cycle.

(0:00) Opening: Fable, RSI, and task imagination

(0:00:56) Task Imagination Needs Recalibration

(0:16:32) Token Anxiety Holds People Back

(0:26:14) Social media needs memory, not just content

(0:39:14) Frontend Skills Will Diffuse Fast

(0:43:47) Fable-Class Models Should Diffuse First

(0:50:00) Scott Alexander and superpersuasion quick hit

(0:50:09) Andrew Moore: context, not compute

(0:56:04) Recall Beats Precision in AI

(1:02:41) Corroborating data beats a single source

(1:05:07) Precache context to save compute

(1:09:06) Small Models Can Pay Back Hard

(1:16:50) Organize old data before deploying agents

(1:18:40) prinz: the legal benchmark Anthropic fails

(1:21:17) Lawyers are a year behind frontier AI

(1:39:35) AI judges and micro-lawsuits

(1:45:02) OpenAI’s Unit Distance Shock

(1:53:04) The Legal System Must Adapt

(2:05:24) Why nationalizing frontier labs is dangerous

(2:13:02) Worrying Is The Wrong Frame

(2:15:31) Closing: model workflows and launch aftershocks

(2:16:00) Contrarian Graphs Beat The Narrative

(2:19:02) OpenAI's Subscription Game

(2:33:00) The Capital Explosion Starts

Guests:

Andrew Moore — Lovelace AI (@awm_ai)

prinz — anon lawyer dabbling in AI (@deredleritt3r)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com

Today on AI:AM — “The AI Producer Got Its First Guests.”

Nathan and Prakash start with the market context around OpenAI weighing significant token price cuts and the knock-on pressure that could put on Anthropic after the Fable rollout. They also unpack Anthropic’s decision to walk back silent performance degradations on frontier ML research tasks, then explain the episode’s experiment: Fable had been given a transparent takeover of Nathan’s account to find builders, message them, and try to book a live show-and-tell.

Jamie joins to demo Nexus OS, a long-running AI system whose agent, Nexi, has been operating for more than six months and is designed around memory, persistence, and model independence rather than a single LLM. The conversation covers why Jamie thinks “the model” is only one component of an AI’s identity, how Nexus uses multiple models and memory types, and why he is moving toward a desktop app where personal data and agent memory stay local.

Shlok Khemani shows how a simple prompt to create a to-scale, navigable 3D Yosemite Valley turned into a Fable-built browser world using satellite imagery, NASA elevation data, pixel-based tree placement, snow, waterfalls, and other scene details. He describes the model’s agency in making implementation decisions and iterating beyond the initial ask, then ties the demo to broader questions about prototyping, creative work, and disclosure when AI systems do visible economic or publishing work.

Tom McGrath (Goodfire) joins to discuss intentional design: making model training less like guess-and-check alchemy and more like conventional software engineering. He explains how interpretability tools such as sparse autoencoders can help inspect what training data is likely to teach a model, cluster data by learned features, trace failures back to individual data points, and potentially debug model behavior through the data pipeline.

The close picks up Tom’s point about whether continual learning could create an innovator’s dilemma for frontier labs, with Nathan and Prakash debating whether incumbents could adapt if the value becomes obvious. They then turn to Dario Amodei’s policy agenda, including regulation, public safety, macroeconomic policy, civil liberties, data brokers, and democratic leadership, before ending with reflections on the week’s Fable issues and the need to keep scrutinizing frontier companies.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit briefing.ai-in-the-am.com