The AI Chefs: The Department of Defense vs. Anthropic's "Monster Robots"

Welcome back to the AI Kitchen! In this episode, Chef Kevin Kieller and Chef David Maldow dish out the latest AI news, starting with a major honor: Chef David has been named a Zoom MVP! The Chefs then dive into the high-stakes political battle surrounding frontier AI models. They discuss OpenAI’s limited preview of GPT-5.6 (code-named "Sol") and the U.S. government's confusing stance on Anthropic's "Mythos" model. Chef David breaks down how the Department of Defense clashed with Anthropic over using Claude for military applications, prompting California Governor Gavin Newsom to publicly throw his support behind Anthropic. The menu takes a philosophical turn as the Chefs explore the limits of Large Language Models (LLMs). They discuss Meta’s former Head of AI arguing that we need "World Models" grounded in physical reality to achieve true Artificial General Intelligence (AGI). They also analyze a new scientific study revealing that humans inadvertently train AI to be "sycophants," upvoting AI responses that simply agree with us rather than tell the truth. Finally, the Chefs look at OpenAI's new partnership with Broadcom to develop their own AI processors (hilariously code-named "Jalapeño") to escape the "Nvidia tax," and announce the upcoming in-person AI Chefs Lab at the CEDIA/Commercial Integrator Expo in Denver. Hosted by David Maldow of Let's Do Video and Kevin Kieller. Chapters: 0:23 Welcome & Heat Dome Weather 2:05 Chef David is Named a Zoom MVP! 4:11 OpenAI Previews GPT-5.6 ("Sol") 5:01 The US Government's Confusing Stance on Anthropic 7:14 California Governor Backs Anthropic Over OpenAI 11:15 Moving Beyond LLMs to Physical "World Models" 15:21 Study: AI is Trained to be a Sycophant 19:09 Chef David's Fight with Claude's False Positives 22:01 OpenAI Develops "Jalapeño" Chips to Escape the Nvidia Tax 24:50 In-Person Event Announcement: CEDIA Expo in Denver 26:41 Wrap Up & Microsoft Copilot Co-work Billing Warning