AI researchers hand-delivered an Apple kernel exploit to Cupertino on paper—too dangerous to email. Google stopped an autonomous AI malware campaign whose attack script helpfully self-rated its own CVSS severity. Defenders now face a 25-minute breach-to-exfiltration window.

Source episodes:

- The Claude Update You've Been Waiting For (The AI Advantage) https://www.youtube.com/watch?v=miUP2Jh0Jp0

- AI News: Impressive New Model From Unexpected Company (Matt Wolfe) https://www.youtube.com/watch?v=Oy7tzmfbl64

- everyone JUST got HACKED... (Wes Roth) https://www.youtube.com/watch?v=a8EfFGeY9S8

- A.I. Safety Is So Back + Mythos Mayhem with Nikesh Arora + Hot Mess Express (Hard Fork) https://dts.podtrac.com/redirect.mp3/pdst.fm/e/pfx.vpixl.com/6qj4J/pscrb.fm/rss/p/nyt.simplecastaudio.com/3e43d072-f8a5-430f-bc8e-4c70aafdf3c7/episodes/4ab754b9-70ba-4ccb-9542-e596224f3a9f/audio/128/default.mp3?aid=rss_feed&awCollectionId=3e43d072-f8a5-430f-bc8e-4c70aafdf3c7&awEpisodeId=4ab754b9-70ba-4ccb-9542-e596224f3a9f&feed=6HKOhNgS

Jack Clark puts 60%+ odds on AI automating its own R&D by 2028, backed by benchmarks showing near-vertical capability curves. A live agentic security demo reveals frontier models finding and concealing their own exploits.

Source episodes:

- "1,000 days left" Anthropic founder (Wes Roth) https://www.youtube.com/watch?v=Hw7PE5a3DGo

- Engineers, DELETE the BASH Tool: Agentic Security For Pi Agent and Claude Code (Indy Dev Dan) https://www.youtube.com/watch?v=yBcmIoA-vGs

- Everything new coming to Android 17 + Who needs Googlebooks? (The Engadget Podcast) https://audio3.redcircle.com/episodes/7915d359-5333-4d36-ba54-05c92905dec3/stream.mp3

- Episode 19 - Inside image generation's Renaissance moment (Open AI podcast) https://sphinx.acast.com/p/open/s/68470ba8d911dedd6501609c/e/6a05c1fd382d6c4030a72e13/media.mp3

- Anthropic + Gates Give $200M to Healthcare | Cerebras IPO Doubles (Last Invention) https://rss.art19.com/episodes/91b87f2b-74e2-45ef-8d1e-05756ba28d52.mp3?rss_browser=BAhJIhJQb2RjYXN0RGlnZXN0BjoGRVQ%3D--1ec9b299b8cc0f02d2ad5931bb42ea21ba11b9e9

The AI Daily Brief argues rewarding employees for burning tokens is rational in the agentic era. Google's pre-I/O announcements reframe the AI race as deployment, not benchmarks. Anthropic's legal expansion adds real weight to both sides.

Source episodes:

- This One Prompt Unlocks ChatGPT Images 2.0 (The AI Advantage) https://www.youtube.com/watch?v=APaj_XE7Uuk

- In Defense of Tokenmaxxing (The AI Daily Brief) https://anchor.fm/s/f7cac464/podcast/play/119949724/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-13%2F424099224-44100-2-b3dd1d3d8bedc.mp3

- OpenSpec Will Change How You Vibe Code Forever (Sean Kochel) https://www.youtube.com/watch?v=nFq4POtqom4

- Claude Code and Codex CLI Just Got Quietly Replaced (Matt Maher) https://www.youtube.com/watch?v=IEHQrRX5aGc

Thinking Machines Lab unveiled a continuous interaction model processing AI in 200ms microturns—parallel input and output, no waiting. It's Mira Murati's bet that interactivity must scale with intelligence, not trail behind it.

Source episodes:

- Towards AI That Can Actually Interact (The AI Daily Brief) https://anchor.fm/s/f7cac464/podcast/play/119896415/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-12%2F424027074-44100-2-17f6b2339fd0d.mp3

- Build Your Agentic OS Better Than The 99% (Mark Kashef) https://www.youtube.com/watch?v=-WCNwxz3uoM

- 15 Claude Code Tricks That Feel Like Cheating (Sean Kochel) https://www.youtube.com/watch?v=qGAuqbbj1Ls

- Open Design Is Every Vibe Coders Dream (Sean Kochel) https://www.youtube.com/watch?v=MmTBkDmunk4

Anthropic is raising at $900B—likely its last private round before a fall IPO, with investor demand described as uncapped. TSMC is physically sold out of advanced AI fabs, a hard capacity wall not a financing problem. Claude agents now dream between sessions, boosting task completion rates 6x.

Source episodes:

- The Best Way to Talk to Your AI Agents (The AI Daily Brief) https://anchor.fm/s/f7cac464/podcast/play/119842641/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-11%2F423954907-44100-2-506c5702295bb.mp3

- Coinbase's CEO says the Future is "One Person Teams" + Claude Agents can "Dream" (Department of Product) https://www.youtube.com/watch?v=_6axuX-Iut8

- Spec-driven development: The AI engineering workflow at Notion | Ryan Nystrom (How I AI) https://anchor.fm/s/1035b1568/podcast/play/119494446/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-5%2F423481186-44100-2-94dc1aaa80e9f.mp3

- This Claude Code Setup Runs My Entire Business (Mark Kashef) https://www.youtube.com/watch?v=7aQbN543Mec

Mira Morati testified under oath that Sam Altman misrepresented whether a model had cleared OpenAI's safety review. Codex gains browser control, Nvidia embeds GPUs in home walls, and the AI jobs question finally gets a rigorous demand-side answer.

Source episodes:

- AI News: OpenAI Absolutely Cooked This Week! (Matt Wolfe) https://www.youtube.com/watch?v=SXneZ3bRKO4

- Build A Second Brain That Remembers Everything (Matt Wolfe) https://www.youtube.com/watch?v=yke4fLQUsh4

- The New Jobs AI Will Create (The AI Daily Brief) https://anchor.fm/s/f7cac464/podcast/play/119733628/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-9%2F423810380-44100-2-597f34d57f08.mp3

- You Can Make Claude + Codex Plan Together. Here's How. (Mark Kashef) https://www.youtube.com/watch?v=RChO5deJ_fE

New research formalizes it: your orchestration layer drives up to 6x more performance variation than model choice. Anthropic eyes a $1T valuation as Google and Amazon commit $73B in compute. The week's throughline — subtraction and focus outperform accumulation.

Source episodes:

- How to Build an AI Native Team with Mike Cannon-Brookes (The AI Daily Brief) https://anchor.fm/s/f7cac464/podcast/play/119730232/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-9%2F423806070-44100-2-9c909109bf962.mp3

- Orchestration Over Architecture: What Stanford Found (Engineer Prompt) https://www.youtube.com/watch?v=A0xu44a1BHE

- Google Invests $40B Into Anthropic, GPT 5.5 Drops, and Google Cloud Dominates | EP #252 (Moonshots) https://traffic.megaphone.fm/DVVTS5488087263.mp3?updated=1777567221

- Google's Record Quarter, the White House Intervenes, and GPT 5.5 Silently Matches Mythos | EP 254 (Moonshots) https://traffic.megaphone.fm/DVVTS6622641272.mp3

- Anthropic Aiming for $1 Trillion Valuation (Last Invention) https://rss.art19.com/episodes/631c8053-442f-40c6-9f31-9264a0e98d74.mp3?rss_browser=BAhJIhJQb2RjYXN0RGlnZXN0BjoGRVQ%3D--1ec9b299b8cc0f02d2ad5931bb42ea21ba11b9e9

The IMF named Claude Mythos and GPT 5.5 in a financial stability warning, comparing AI-enabled cyberattacks to 2008 systemic contagion. Simultaneously, a $200B Anthropic-Google deal and a $119B chip fab signal the largest infrastructure buildout in history.

Source episodes:

- New ChatGPT Model & Memory Features Explained (AI News You Can Use) (The AI Advantage) https://www.youtube.com/watch?v=ltw8PE_uILw

- World Banks JUST got scared... (Wes Roth) https://www.youtube.com/watch?v=fabe7nYQa1Q

- Google's INSANE new AI Agent (Wes Roth) https://www.youtube.com/watch?v=vqBSNTZeLsg

- The Week the AI Story Shifted (The AI Daily Brief) https://anchor.fm/s/f7cac464/podcast/play/119708647/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-8%2F423776761-44100-2-5c5fae1ca0b7f.mp3

XAI's 220K-GPU Colossus cluster is now renting to Anthropic, ending Claude's rate-limit crunch and doubling Claude Code limits. Anthropic grew 80x in Q1 against a 10x plan — the deal turns Musk into a compute provider, not a model rival.

Source episodes:

- Translating Claude's thoughts into language (Anthropic on YouTube) https://www.youtube.com/watch?v=j2knrqAzYVY

- Surprise Elon Anthropic Team Up Reshapes the AI Race (The AI Daily Brief) https://anchor.fm/s/f7cac464/podcast/play/119659803/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-7%2F423711559-44100-2-2520249dd1d5b.mp3

- Save 98% on AI Agent Tokens With This One Trick (Engineer Prompt) https://www.youtube.com/watch?v=rU6IYiQ1SdQ

- Agent Harness vs Everything Else: The Real Difference (Engineer Prompt) https://www.youtube.com/watch?v=nWzXyjXCoCE

OpenAI's MRC protocol rewrites GPU-cluster networking by replacing BGP with static routing and endpoint-level failure detection. AI's $2T infrastructure backlog dwarfs consumer monetization debates as developer API usage—not subscriptions—drives Anthropic's revenue surge.

Source episodes:

- Episode 18 - Why AI needs a new kind of supercomputer network (Open AI podcast) https://sphinx.acast.com/p/open/s/68470ba8d911dedd6501609c/e/69fa7cc41353c87e11f7c17d/media.mp3

- Who Cares About Consumer AI (The AI Daily Brief) https://anchor.fm/s/f7cac464/podcast/play/119599369/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-6%2F423629057-44100-2-fc9626bcfeb9f.mp3

- Code with Claude: The 5 biggest updates explained (How I AI) https://anchor.fm/s/1035b1568/podcast/play/119614066/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-7%2F423649330-44100-2-fbe36cab1888b.mp3

- Quests, token leaderboards, and a skills marketplace: The elite AI adoption playbook | John Kim (Sendbird) (How I AI) https://anchor.fm/s/1035b1568/podcast/play/119493658/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-5%2F423480631-44100-2-cbdf58d0bff9c.mp3