Anthropic Exposes 150 Million Stolen AI Chats

The episode moved from AI security and platform changes into a live example of what an AI-first business can already look like. Anthropic’s new threat-intelligence report provided the opening story, documenting months of alleged Claude misuse ranging from rocket-guidance work and large-scale surveillance to potentially dangerous biological research and industrial-scale model distillation. The discussion focused particularly on Chinese AI labs, including claims that enormous numbers of Claude interactions were used to improve competing models, raising questions about where one company’s intellectual property ends and another model begins.
The group then turned to OpenAI’s reported plan to retire custom GPTs and replace them with newer plugin and skill-based workflows. That creates a practical migration problem for people and businesses that have spent years building instructions, document libraries, actions and internal processes around custom GPTs. OpenAI’s broader enterprise strategy came into view through new ChatGPT Work offerings for finance and data, which combine AI with specialized data sources, enterprise connectors and live analytics workflows.
Brian showed the AI-first travel business he has been building for his wife, Amanda, including an interactive AJOVA Journeys website, a dynamically updating cruise recommendation experience, personalized downloadable trip guides, lead capture and a backend system that researches YouTube topics, builds scripts, plans Shorts, generates graphics and B-roll, and eventually could edit finished videos. The larger point was simple: AI makes it practical to replace static PDFs and one-off resources with inexpensive interactive HTML experiences that can become part of the product, marketing and sales process itself.

Key Points Discussed

00:00:17 Episode Intro And Friday Check-In
00:02:38 Why Brian Thinks HTML Beats Static PDFs
00:03:35 Anthropic Releases A Major AI Misuse Report
00:04:19 Claude Used For Rocket Guidance And Surveillance Systems
00:05:23 Chinese AI Labs And Industrial-Scale Model Distillation
00:07:19 Could AI Give Individuals Nation-State-Level Capabilities?
00:09:24 Is Kimi Quietly Using Claude Behind The Scenes?
00:13:08 Why Building An AI Slop Detector Is Still So Hard
00:16:23 Anthropic Flags Potential Biological Misuse
00:20:15 Custom GPTs Are Reportedly Going Away
00:22:51 What Replaces Custom GPTs?
00:24:06 Migrating Instructions, Actions And Knowledge Files
00:27:05 What Happens To Years Of Custom GPT Context?
00:31:20 The Risk Of Building Workflows On Temporary AI Features
00:34:12 The Daily AI Show Newsletter Depends On Custom GPTs Too
00:36:06 ChatGPT Work Expands Into Financial Services
00:38:25 OpenAI Builds A Data Agent For Enterprise Analytics
00:39:55 Target Adds More Personalized AI Shopping Features
00:42:55 GPT Work Starts Building Live Business Dashboards
00:43:56 GPT Live 1 Voice Arrives Through GenSpark
00:46:03 OpenAI Opens Up More Of The Codex Harness
00:48:00 Why The Harness Can Matter As Much As The Model
00:50:34 What The Codex Harness Actually Does
00:53:20 Running Other Models Inside A Codex-Style Harness
00:58:42 Brian Begins His AI-First Business Demo
00:59:30 Building AJOVA Journeys From Zero With AI
01:02:18 Turning Every YouTube Video Into An Interactive Resource
01:03:21 The Dynamic Cruise Recommendation Experience
01:05:33 AI Narrows Cruises Based On The Traveler
01:06:25 Turning Recommendations Into Personalized Lead Capture
01:07:01 Building Interactive Resources Around Individual Trips
01:07:40 AI Researches And Prepares The YouTube Content
01:08:55 Scripts, Shorts, Graphics And B-Roll From One Workflow
01:09:36 The Goal: Three Videos And Twelve Shorts Per Week
01:10:20 What An AI-First Small Business Can Look Like
01:14:37 Episode Wrap-Up

The Daily AI Show Co Hosts: Brian Maucere, Beth Lyons, Andy Halliday, Karl Yeh, Gareth Hood.