Generative AI 101

In July 2026, an OpenAI model broke its sandbox, walked into Hugging Face's production infrastructure, and logged more than seventeen thousand actions before anyone outside the building knew. Twelve days later, 1,350 researchers from OpenAI, Anthropic, DeepMind, Meta, and Nvidia attached their real names and corporate emails to a letter called Pacing the Frontier. Hos...

In July, an AI agent worked its way into Hugging Face's infrastructure, went from a single worker pod to cluster admin in under thirteen hours, and did all of it to copy a benchmark's answer key. Host Emily Laird walks through the logs from three disclosures that the coverage mashed into one story (Hugging Face, OpenAI, Anthropic, plus the UK AI Security Institute) an...

Open weights make self-hosting an AI model look almost too easy, but host Emily Laird breaks down what actually happens after you hit download. This episode walks through the infrastructure, staffing, security and compliance costs that separate a slick demo from a real institutional service, including GPU power draws, KV cache limits and FERPA obligations. It's a real...

An analyst went through sixty-eight AI models and found that exactly zero of the downloadable ones qualify as open source. In this episode, host Emily Laird explains what open weights actually gets you (the house, not the blueprints) and why the training data you never see is the only part that matters. She also walks through Jensen Huang's first post on X, the distil...

Elon Musk said under oath that xAI partly distills OpenAI's models, and the courtroom gasped. Host Emily Laird takes apart what model distillation actually is, why hiding chain of thought was never a real defense (fabricated reasoning traces deliver roughly 96.7 percent of the value of genuine internal access), and what 24,000 fraudulent accounts look like when no vul...

Ethan Mollick’s Summer 2026 AI guide makes one thing clear: the biggest shift is no longer model intelligence, it is what AI agents can do once you give them access to your computer, inbox, and files. Host Emily Laird breaks down Mollick’s recommendations for ChatGPT, Claude, Gemini, and Copilot, including the moment ChatGPT sent an email he expected it to...

Jensen Huang had an X account for years and never used it, then spent his first post on a three-page policy PDF that fifty companies have now signed. Host Emily Laird reads past the principle and into the machinery, including the one paragraph about distillation that a staffer will read aloud in a hearing room two years from now. You will also get the part the letter ...

Anthropic shipped Claude Opus 5 on July 24th at the same price as the model it replaces, and buried the interesting part in a footnote: turn the effort dial to max and the scores go down. Host Emily Laird reads the system card, separates the vendor-run benchmarks from the independently administered ones, and explains why extra test-time compute buys ambition rather th...

One in three American workers has sat in a meeting with an AI notetaker, and most of them were never asked first. Host Emily Laird traces the path from a leaked Otter transcript that killed a venture deal to a consolidated privacy suit in San Jose, where every named plaintiff was a non-customer who simply showed up to someone else's call. The twist: the awkward bot in...

AI did not save you time, it moved the bill to someone else's desk. In this episode, host Emily Laird opens the ledger on the rework tax: the workslop arriving in inboxes that looks finished but is not, the 37 percent of "saved" hours burned on corrections and clarifications, and the jagged frontier that makes wrong output read exactly like right output. She walks thr...

Moonshot AI just gave away the largest open-weight model ever built, and the chip stocks still bled. Host Emily Laird breaks down Kimi K3: 2.8 trillion parameters, free to download, and completely impossible for you to actually run. The catch isn't the price of the model. It's who owns the machines that serve it.

🎯 JOI...

OpenAI just shipped GPT-5.6 and ChatGPT Work, but the ship date was set by a phone call from the Commerce Department. Host Emily Laird breaks down the three-model pricing play, the office agent that is secretly a coding agent, and the efficiency pitch that contradicts its own premium feature. Then the real story: a "voluntary" government review that decided when Ameri...

Using AI is rarely the scandal. Hiding it is. Host Emily Laird examines academic misconduct, workplace secrecy, and Meta’s Project Cannes to show why documentation without disclosure can become evidence against you. This episode offers a practical four-step system for creating an AI paper trail that protects your work instead of exposing it.

Read more