# Anthropic to Raise at $900B, Musk Admits Grok Was Trained on OpenAI, Gemini Hits 4M Cars Page: https://stenobird.com/podcast/chatgpt-news/anthropic-to-raise-at-900b-musk-admits-grok-was-trained-on-openai-gemini-hits-4m-cars Text version: https://stenobird.com/podcast/chatgpt-news/anthropic-to-raise-at-900b-musk-admits-grok-was-trained-on-openai-gemini-hits-4m-cars.md Podcast: [ChatGPT: News on Open AI, MidJourney, NVIDIA, Anthropic, Open Source LLMs, Machine Learning](https://stenobird.com/podcast/chatgpt-news) Published: 2026-05-01T20:43:50+00:00 Episode link: https://rss.art19.com/episodes/f96c0a37-ca3d-4da7-b08e-279649192970.mp3?rss_browser=BAhJIg90cmFuc2NyaWJyBjoGRVQ%3D--952c5701c84ad333c69d5faa668f8177091704f0 Audio file: https://rss.art19.com/episodes/f96c0a37-ca3d-4da7-b08e-279649192970.mp3?rss_browser=BAhJIg90cmFuc2NyaWJyBjoGRVQ%3D--952c5701c84ad333c69d5faa668f8177091704f0 Processing state: processed JSON: https://stenobird.com/v1/public/podcasts/chatgpt-news/episodes/anthropic-to-raise-at-900b-musk-admits-grok-was-trained-on-openai-gemini-hits-4m-cars Duration seconds: 741 ## Resource The Pentagon's new approved AI vendor list notably excludes Anthropic, signaling a strategic shift toward alternative frontier labs. Meanwhile, Google expands Gemini's footprint into millions of vehicles and Elon Musk addresses allegations of using OpenAI data to train Grok. ## Highlights - Main idea: The Pentagon has officially authorized seven AI companies for its classified network, conspicuously leaving Anthropic off the list - Strategic shift: The inclusion of smaller, specialized firms like Reflection AI suggests the DOD is actively cultivating alternatives to dominant players - Practical takeaway: Google's Gemini integration into GM and Volvo vehicles marks a massive-scale deployment of LLMs into consumer hardware - Failure mode: The practice of model distillation—using high-quality outputs from one model to train another—remains a controversial but widespread industry tactic - Market insight: Anthropic is reportedly pressuring investors with a 48-hour window to participate in a high-valuation funding round ## Topics Anthropic, Pentagon AI Strategy, Google Gemini, Elon Musk, xAI, Model Distillation, Claude Security, Automotive AI, Defense Technology ## Chapters - 1:00 — Gemini's Automotive Expansion: Google is integrating Gemini into millions of GM and Volvo vehicles, replacing Google Assistant with advanced multimodal capabilities. - 2:40 — Claude Security Beta: Anthropic launches Claude Security, a tool designed to scan codebases for vulnerabilities and recommend automated patches. - 4:30 — Enterprise AI Integration: Major cybersecurity and consulting firms like CrowdStrike and Deloitte are rapidly integrating Claude into their service platforms. - 5:20 — The Grok Training Controversy: Elon Musk addresses legal questions regarding whether xAI utilized OpenAI's models for distillation during Grok's training. - 7:20 — Pentagon's Approved AI List: An analysis of the seven companies, including OpenAI and NVIDIA, granted access to the Pentagon's classified networks. - 8:10 — The Anthropic Exclusion: Discussion on why Anthropic was omitted from the DOD list and the implications of being labeled a supply chain risk. - 10:40 — The Rise of Reflection AI: How the Pentagon is leveraging specialized labs like Reflection AI to build custom, sovereign AI capabilities. ## Actions - request_transcript: `POST https://stenobird.com/v1/public/podcasts/chatgpt-news/episodes/anthropic-to-raise-at-900b-musk-admits-grok-was-trained-on-openai-gemini-hits-4m-cars/transcription-requests` — Idempotently request low-priority transcript generation for this episode. - read_markdown: `GET https://stenobird.com/podcast/chatgpt-news/anthropic-to-raise-at-900b-musk-admits-grok-was-trained-on-openai-gemini-hits-4m-cars.md` — Read the agent-friendly Markdown representation of this episode resource. A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed. ## Transcript Full transcripts are not published on public pages unless there is a clear rights basis.