AI with Kyle

Epizódy

• 
  Anthropic Wants to Slow AI Down. Even Elon Musk Agrees. 14.09.2026 29min

  Get my daily AI newsletter: https://aiwithkyle.com/Anthropic wants to slow AI development down, and Sam Altman, Elon Musk and Demis Hassabis are backing the direction of Dario Amodei's proposal. These are usually fierce rivals. Now they're calling for a slower pace, outside evaluators and shared safety standards. I'm sceptical about what the biggest AI companies stand to gain.I go through Amodei's essay, "We Must Pace the Frontier", and the full posts from the people supporting and challenging it. David Sacks tells the labs to slow down themselves. Cohere's Aidan Gomez calls out a possible cartel. Donald Trump wants America to stay ahead of China.We also look at recursive self-improvement, OpenAI's growing use of coding agents, METR's role as an evaluator, and the problem of verifying an international agreement. AI safety deserves serious attention. So do the costs of delaying useful tools and making it harder for smaller companies to compete.Recorded 14 September 2026.*RESOURCES*Previous video on Jacob Coxon: https://www.youtube.com/watch?v=6gcpXBLmwPESelf-improving AI explained: https://www.youtube.com/watch?v=Os-s26O_W08*CHAPTERS*00:00 AI rivals suddenly agree01:32 Dario, Sam, Musk and Demis02:52 The Jacob Coxon connection04:12 What does AI pacing mean?05:21 Embedded evaluators07:05 Pacing within democracies08:36 Can a global slowdown work?09:54 Recursive self-improvement11:45 David Sacks: then slow down16:19 Aidan Gomez and the cartel accusation18:40 Who checks the checkers? METR21:15 The China problem22:28 Trump's response24:06 Derek Thompson's four-point argument25:19 Slowing down has a cost27:19 Safety and commercial self-interest*KEY POINTS*• Rival AI labs broadly agree on pacing, but the rules are still being debated.• Outside evaluators need access, independence and the ability to publish findings.• Expensive compliance requirements can favour the biggest companies.• Any international slowdown faces the problem of verifying what other countries do.• Delaying useful AI carries costs for medicine, work and smaller businesses.*SOURCES AND TOOLS MENTIONED*Dario Amodei's essay: https://darioamodei.com/post/we-must-pace-the-frontierDario's announcement: https://x.com/DarioAmodei/status/2098773920774074715Sam Altman's full post: https://x.com/sama/status/2099348812305473766Sam's initial response: https://x.com/sama/status/2098811563415150910Elon Musk: https://x.com/elonmusk/status/2098789109980332057Demis Hassabis: https://x.com/demishassabis/status/2098909516582490602David Sacks: https://x.com/DavidSacks/status/2098973625252708460Aidan Gomez: https://x.com/aidangomez/status/2098828132396449939OpenAI's research acceleration data: https://openai.com/index/research-acceleration-view-inside-openai/METR's disclosures: https://metr.org/aboutTrump's original video: https://x.com/RapidResponse47/status/2099155148576542928Derek Thompson: https://x.com/DKThomp/status/2098819497821917480Ethan Mollick: https://x.com/emollick/status/2099134552807522610*LISTEN TO AI WITH KYLE*Spotify: https://open.spotify.com/show/6pnwfIJvxfC8jlX891XBiEApple Podcasts: https://podcasts.apple.com/us/podcast/ai-with-kyle-daily-ai-news-and-updates/id1825779580*CONNECT WITH KYLE*Website: https://aiwithkyle.comTikTok: https://www.tiktok.com/@iamkylebalmerInstagram: https://www.instagram.com/iamkylebalmerLinkedIn: https://www.linkedin.com/in/kylebalmer/#Anthropic #OpenAI #ArtificialIntelligence
• 
  Anthropic Insiders Say AI Could Kill Us All. Are They Right? 11.09.2026 40min

  Want to build with AI but have no idea where to start? https://aiwithkyle.com/join-freedom-osJoin my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chatMedia/Sponsorship Inquiries: https://aiwithkyle.com/media-kitGet my daily AI newsletter: https://aiwithkyle.com/Subscribe and turn on notifications to catch the next live stream:https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgJacob Coxon resigned from Anthropic and accused the leading AI labs of gambling with our lives. Then came the interviews, the donor connections people started digging into, and accusations that the whole thing was a "psyop".I go through Coxon's original warning and the responses from Beff Jezos, Parker Thayer, Elon Musk and Bernie Sanders. We also look at recursive self-improvement, what Anthropic's own researchers have said, and who would get to define and enforce a superintelligence ban.Recorded 11 September 2026.*CHAPTERS*00:00 The resignation and the accusations01:59 What Coxon actually said05:50 Who is Jacob Coxon?07:43 Why people think it was organised14:22 The interviews17:46 The pushback24:04 What recursive self-improvement means25:51 What Anthropic's researchers are saying27:52 Where the proposed solutions overlap29:57 Who benefits from slowing down?34:33 The commercial angle36:57 More than one thing can be true*SOURCES AND RESOURCES*Jacob Coxon's original post:https://x.com/hilbertspaess/status/2097476196791709843Parker Thayer:https://x.com/ParkerThayer/status/2097759699626328575Beff Jezos:https://x.com/beffjezos/status/2097818903321993315Elon Musk on the post's reach:https://x.com/elonmusk/status/2097908044344746406Bernie Sanders:https://x.com/SenSanders/status/2098078825909219721Evan Hubinger:https://x.com/EvanHub/status/2097497037956891126https://x.com/EvanHub/status/2097528891846074828Anthropic on alignment and security:https://www.anthropic.com/news/improving-alignment-security-effortsSanders and Casar's proposed legislation:https://www.sanders.senate.gov/press-releases/news-sanders-casar-introduce-legislation-to-ban-artificial-superintelligence-and-temporarily-pause-advanced-ai-development/Anthropic's policy proposals:https://www.anthropic.com/news/2028-ai-leadership*LISTEN TO AI WITH KYLE*Spotify: https://open.spotify.com/show/6pnwfIJvxfC8jlX891XBiEApple Podcasts: https://podcasts.apple.com/us/podcast/ai-with-kyle-daily-ai-news-and-updates/id1825779580*CONNECT WITH KYLE*TikTok: https://www.tiktok.com/@iamkylebalmerInstagram: https://www.instagram.com/iamkylebalmerLinkedIn: https://www.linkedin.com/in/kylebalmer/
• 
  AI Skills Explained: Stop Repeating Yourself to ChatGPT 10.09.2026 14min

  Want to build a business with AI but have no idea where to start? Build your personalized AI income plan and execute this week: https://aiwithkyle.com/join-freedom-osJoin my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chatMedia/Sponsorship Inquiries: https://aiwithkyle.com/media-kitAI Skills let you save the instructions for work you do regularly, then reuse them in ChatGPT, Codex or Claude. A quarterly report, a client update or a writing check can follow the same process each time, with the right examples and supporting files already attached.I walk through a real SKILL.md file, show what belongs in the instructions, and explain when to add references, visual assets or scripts. You'll see how a skill fits alongside prompts, Projects and other saved context, how to find existing skills, and how to choose a useful first task.You can build that first skill by talking through the job with your AI. Give it a specific task, explain what a good result looks like, and test the finished instructions on real examples. Start with one repeatable job and improve the skill as you use it.*RESOURCES*Get the free AI Skills 101 guide: https://aiwithkyle.com/mini/ai-skills?utm_source=youtube&utm_medium=organic_video&utm_campaign=mini_ai_skills&utm_content=ai_skills_101_recut*CHAPTERS*0:00 What is an AI skill?1:32 Using a skill2:31 Inside SKILL.md4:10 References, assets and scripts4:41 A library of skills7:00 Finding existing skills8:16 Choosing a useful task9:44 Building your own skill12:11 Test and improve it*KEY POINTS*- Put the repeatable process in SKILL.md.- Add references and assets when the task needs them.- Choose one useful job for your first skill.- Ask your AI to interview you about how you do that job.- Test the result and revise the instructions.*SOURCES AND TOOLS MENTIONED*OpenAI - Build Skills: https://learn.chatgpt.com/docs/build-skillsOpenAI Academy - Using Skills: https://openai.com/academy/skills/Agent Skills specification: https://agentskills.io/specificationAnthropic - Agent Skills: https://www.anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills*LISTEN TO AI WITH KYLE*Spotify: https://open.spotify.com/show/6pnwfIJvxfC8jlX891XBiEApple Podcasts: https://podcasts.apple.com/us/podcast/ai-with-kyle-daily-ai-news-and-updates/id1825779580*CONNECT WITH KYLE*Website: https://aiwithkyle.comTikTok: https://www.tiktok.com/@iamkylebalmerInstagram: https://www.instagram.com/iamkylebalmerLinkedIn: https://www.linkedin.com/in/kylebalmer/#AISkills #ChatGPT #Claude
• 
  Make ChatGPT Sound Like You: 6 Methods 09.09.2026 27min

  Get the free AI Writing bundle: writing guide, checklist, prompts and reusable skill: https://aiwithkyle.com/mini/goblinsJoin my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chatBuild your AI second brain with GPT-6 Astra: https://www.youtube.com/watch?v=Fc5PcxMPigkFree AI Brain setup guide: https://aiwithkyle.com/mini/ai-brainBuild your personalised AI income plan with Freedom OS: https://aiwithkyle.com/join-freedom-osFree weekly webinar: https://aiwithkyle.com/webinarMedia/Sponsorship Inquiries: https://aiwithkyle.com/media-kitFree 5-Day AI Challenge: https://aiwithkyle.com/joinChatGPT can learn your writing style from connected apps. I show you where to find the setting and how to choose the sources it uses, then walk through five more ways to make AI writing sound closer to you.We cover your finished writing, voice notes and transcripts, a portable voice guide, my AI Tells editing checklist, and feeding your published edits back into your AI. I also show my own voice guides and explain how this setup fits into an AI brain you can use with ChatGPT, Codex and Claude.Keep the examples you like, read the draft before you publish it, and show your AI the changes you made.Chapters:00:00 Make ChatGPT sound like you00:31 Six methods for your own voice00:49 ChatGPT writing style01:32 Why so much AI writing sounds the same05:00 Where to find the writing-style setting07:08 Your finished writing09:11 Voice notes and transcripts11:24 Build a voice guide13:03 My own voice guide examples17:10 The AI Tells checklist19:45 Let AI learn from your final edits23:01 The six methods together24:27 Keep it all in your AI brain26:20 Free resources and next stepsResources:AI Tells skill: https://github.com/kdgbalmer/ai-tellsOpenAI's writing-style announcement: https://x.com/ChatGPT/status/2097018264048251309Wikipedia's Signs of AI Writing: https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writingNewsletter: https://aiwithkyle.com/Listen to AI with Kyle:Spotify: https://open.spotify.com/show/6pnwfIJvxfC8jlX891XBiEApple Podcasts: https://podcasts.apple.com/us/podcast/ai-with-kyle-daily-ai-news-and-updates/id1825779580Connect with Kyle:Website: https://aiwithkyle.comTikTok: https://www.tiktok.com/@iamkylebalmerInstagram: https://www.instagram.com/iamkylebalmerLinkedIn: https://www.linkedin.com/in/kylebalmer/
• 
  Build Your AI Second Brain With GPT-6 Astra 07.09.2026 31min

  Get the free AI second brain setup guide + copy-and-paste prompts: https://aiwithkyle.com/mini/ai-brainJoin my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chatWant to build a business with AI but have no idea where to start? Build your personalized AI income plan and execute this week: https://aiwithkyle.com/join-freedom-osMedia/Sponsorship Inquiries: https://aiwithkyle.com/media-kitGet AI-Ready with Kyle's free 5-Day Challenge: https://aiwithkyle.com/joinSubscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgYou can start your AI second brain with a plain folder and a conversation. I explain how to use GPT-6 Astra in Codex to set it up, interview you and organise what you tell it into useful context about your work.We go through what you want to be, do and have, how to talk through your ideas, and how ChatGPT/Codex and Claude Code can work from the same saved notes. You give each tool access to the same folder and ask it to read and update that context. Each app still has its own chat history and permissions.I also show my Obsidian brain and explain the optional additions: connected apps, scheduled maintenance, computer use and access from your phone. Start with the computer you already have and add things as you need them.Chapters:00:00 Start your AI second brain01:40 What an AI second brain actually is03:05 It starts with a folder04:26 Ask AI to build the setup05:31 The starter prompt06:57 Let AI interview you07:42 What you want to be, do and have08:58 Talk to your brain10:02 Add useful documents10:43 Obsidian is optional13:01 Use the same brain with different AI tools14:29 Choose a model for the job16:06 Shared notes and nightly maintenance17:20 Connect your everyday apps19:49 Give AI computer access22:28 Teach AI a workflow by showing it24:18 An always-on computer27:13 Use your brain from your phone28:33 The whole system explained30:09 Start small and get the free guideUseful resources:OpenAI Codex: https://chatgpt.com/codex/Claude Code: https://claude.com/product/claude-codeObsidian: https://obsidian.md/Get the written guides and newsletter at https://aiwithkyle.com/Listen to AI with Kyle:Spotify: https://open.spotify.com/show/6pnwfIJvxfC8jlX891XBiEApple Podcasts: https://podcasts.apple.com/us/podcast/ai-with-kyle-daily-ai-news-and-updates/id1825779580Connect with Kyle:Website: https://aiwithkyle.comTikTok: https://www.tiktok.com/@iamkylebalmerInstagram: https://www.instagram.com/iamkylebalmerLinkedIn: https://www.linkedin.com/in/kylebalmer/
• 
  You’ve Got GPT-6 Astra. Now What? 05.09.2026 22min

  Get the free GPT-6 Astra playbook and all six prompts: https://aiwithkyle.com/astrajobs*Want to build a business with AI but have no idea where to start? Build your personalized AI income plan and execute this week: https://aiwithkyle.com/join-freedom-os*Join my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chat*Media/Sponsorship Inquiries: https://aiwithkyle.com/media-kitGet AI-Ready with Kyle's 5-Day Challenge: https://aiwithkyle.com/joinSubscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgSummary:You've just got access to GPT-6 Astra in ChatGPT. Now what? The 3D worlds, video games and ridiculous renders are very cool. But unless you're trying to make GTA 6, what do you actually do with it? I've put together four practical briefs for your health and fitness, business, money and life plan. These are projects you can give ChatGPT Astra using your own files, calendar and goals.I start with a voice conversation to work out what you actually want to tackle. Then I show you how to brief Astra like a capable outside team, with enough context, checkpoints and a clear definition of done. The examples cover a personalised fitness app, a plan to grow owner profit without adding hours, a household finance model and three possible futures for your life. I also cover Ethan Mollick's five-day personal wiki project and the permissions I'd set before letting AI agents get to work. Pick one big job, gather the files and get it moving. Check the outputs and keep consequential decisions under your control. Get qualified advice for medical or financial decisions.—— Time Stamps ——0:00 You've Got Astra. Now What?1:19 Four Projects for Your Life2:32 Start With a Voice Conversation4:09 Be, Do, Have5:02 Find a Project Worth Doing6:39 Brief Astra Like an Outside Team8:25 Ethan Mollick's Five-Day Wiki9:22 Build Your Health and Fitness System11:17 Grow Your Business Without Adding Hours13:30 Organise Your Money15:19 Build Your Life Plan18:43 Set the Permission Boundaries20:37 Pick One Big Job— Useful Resources ——Ethan Mollick on briefing AI like an outside team: https://x.com/emollick/status/2095645030941630763Ethan Mollick's personal wiki project: https://x.com/emollick/status/2095606622055760159Matt Shumer's Manhattan render: https://x.com/mattshumer_/status/2095609734845927525OpenAI's guide to long-running work: https://learn.chatgpt.com/docs/long-running-workFind everything else at https://aiwithkyle.com/*LISTEN TO AI WITH KYLE*Spotify: https://open.spotify.com/show/6pnwfIJvxfC8jlX891XBiEApple Podcasts: https://podcasts.apple.com/us/podcast/ai-with-kyle-daily-ai-news-and-updates/id1825779580*CONNECT WITH KYLE*Website: https://aiwithkyle.comTikTok: https://www.tiktok.com/@iamkylebalmerInstagram: https://www.instagram.com/iamkylebalmerLinkedIn: https://www.linkedin.com/in/kylebalmer/
• 
  Did OpenAI Just Achieve AGI? ChatGPT GPT-6 Astra Explained 04.09.2026 19min

  *Join my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chat*Want to build a business with AI but have no idea where to start? Build your personalized AI income plan and execute this week: https://aiwithkyle.com/join-freedom-os*Media/Sponsorship Inquiries: https://aiwithkyle.com/media-kit===Get AI-Ready with Kyle's free 5-Day Challenge: https://aiwithkyle.com/joinSubscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgGPT-6 Astra launched overnight with one of the wildest benchmark tables I've seen. Its 99.9% ARC-AGI-3 score has people claiming AGI is here. The test measures whether an agent can explore an unfamiliar world, learn its rules and act efficiently. Astra's biggest practical jump is computer use.ARC Prize's standard harness gave Astra 62.7%. OpenAI's Provider Adapter produced the 99.9% result by preserving reasoning state and compacting long conversations. The video also covers a week-long Manhattan build in Unreal Engine, Apple Park in Blender and an interactive Library of Alexandria.I also cover the messy rollout, who gets access first and the banked Codex resets OpenAI is giving paid ChatGPT users while they wait. If Astra is in your account, give it a large computer-use or coding job. An email rewrite will tell you almost nothing about this model.Chapters:0:00 GPT-6 Astra launches0:43 What OpenAI actually changed1:32 Is Astra AGI?2:25 The 99.9% ARC-AGI-3 result3:14 What ARC-AGI-3 actually tests9:19 Why the AGI label remains unsettled9:39 Independent benchmark checks11:08 Astra's practical computer-use jump11:38 Manhattan, Apple Park and Alexandria15:27 Who gets access and when16:47 My verdict on GPT-6 Astra17:59 Final summaryUseful resources:OpenAI's GPT-6 Astra launch page: https://openai.com/index/gpt-6-astra/OpenAI API model page: https://developers.openai.com/api/docs/models/gpt-6-astraOpenAI system card: https://deploymentsafety.openai.com/gpt-6-astraARC Prize results and harness comparison: https://arcprize.org/results/openai-gpt-6-astraSam Altman on the rollout: https://x.com/sama/status/2095678759651438887Artificial Analysis on coding efficiency: https://x.com/ArtificialAnlys/status/2095595494081024077Ethan Mollick's interactive Alexandria example: https://x.com/emollick/status/2095601539066777748Matt Shumer's Manhattan in Unreal Engine: https://x.com/mattshumer_/status/2095609734845927525Find everything else at https://aiwithkyle.com/
• 
  Fable 5.1 Is Built for Bigger Jobs 02.09.2026 22min

  *Want to build a business with AI but have no idea where to start? Build your personalized AI income plan and execute this week: https://aiwithkyle.com/join-freedom-os*Join my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chat*Media/Sponsorship Inquiries: https://aiwithkyle.com/media-kitSubscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgFable 5.1 is here. Anthropic says it is its strongest model for coding and knowledge work, and the early benchmarks are huge. Most people will still use it to rewrite an email. I look at what changed, where the gains show up and how much the extra thinking costs.Give it a bigger job. I show you how to move from asking for an artefact to asking for an outcome, build a useful context pack, set safety boundaries and use Fable like an AI agent across a multi-step project.Chapters:00:00 Fable 5.1 is live01:27 Fable 5.1 benchmarks03:11 Intelligence, price and the independent results04:10 Three ways to cut the cost06:10 The capability overhang08:38 Move from outputs to outcomes11:06 Hand it the whole problem12:50 Four jobs worth giving Fable16:02 Make it finish the whole task16:31 How big AI jobs actually run18:44 One person can borrow a team20:42 Give Fable one big jobUseful resources:Anthropic Fable 5.1 benchmarks: https://www.anthropic.com/claude/fable#benchmarksArtificial Analysis review: https://artificialanalysis.ai/articles/claude-fable-5-1Fable 5.1 prompting guide: https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-1OpenAI on the capability overhang: https://openai.com/index/how-countries-can-end-the-capability-overhang/OpenAI on AI across occupations: https://openai.com/index/how-ai-is-expanding-what-people-do-at-work/Find everything else at https://aiwithkyle.com/
• 
  OpenAI’s AI Chip vs Nvidia’s $13B Hugging Face Bet 31.08.2026 17min

  *Want to build a business with AI but have no idea where to start? Build your personalized AI income plan and execute this week: https://aiwithkyle.com/join-freedom-os*Join my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chat*Media/Sponsorship Inquiries: https://aiwithkyle.com/media-kitOpenAI has published the first results from its Jalapeño inference chip. Nvidia reportedly wants to buy Hugging Face for $12.9 billion. Both companies are trying to depend less on the other.This week's AI news also covers three ChatGPT updates that make agentic AI much easier to use, plus Microduck: a $399 open-source robot that can walk, carry things and recover after a fall.*RESOURCES*ChatGPT Work mini guide: https://aiwithkyle.com/mini/chatgpt-workMy ChatGPT Work walkthrough: https://youtu.be/KLS6i7Sre8o*CHAPTERS*0:00 OpenAI and Nvidia0:39 OpenAI's Jalapeño chip4:29 Nvidia, Hugging Face and local AI10:24 Three ChatGPT updates14:18 The $399 robot duck*KEY POINTS*• OpenAI says its first Jalapeño results beat Nvidia's GB300 on several inference measures.• Jalapeño handles inference. OpenAI still uses Nvidia and other suppliers, especially for training.• Nvidia's reported $12.9 billion Hugging Face deal remains unconfirmed.• ChatGPT Work can sign into websites and react to changes in Gmail, Slack and GitHub.• Microduck is a $399 open-source robot that can learn new movements in a simulator.*SOURCES AND TOOLS MENTIONED*OpenAI's Jalapeño results: https://openai.com/index/jalapeno-first-results/The reported Nvidia-Hugging Face deal: https://www.theinformation.com/articles/nvidia-agrees-buy-open-source-model-repository-hugging-face-12-9-billionChatGPT Work browser update: https://x.com/ChatGPT/status/2092366556584153249ChatGPT Work task update: https://x.com/ChatGPT/status/2092335329110004140Microduck: https://pollen-robotics.com/microduck/blog/introducing-microduck/*LISTEN TO AI WITH KYLE*Spotify: https://open.spotify.com/show/6pnwfIJvxfC8jlX891XBiEApple Podcasts: https://podcasts.apple.com/us/podcast/ai-with-kyle-daily-ai-news-and-updates/id1825779580*CONNECT WITH KYLE*Website: https://aiwithkyle.comTikTok: https://www.tiktok.com/@iamkylebalmerInstagram: https://www.instagram.com/iamkylebalmerLinkedIn: https://www.linkedin.com/in/kylebalmer/#AINews #OpenAI #Nvidia
• 
  AI Wiping Out Graduate Jobs First (Down 46%) - the future of work is rough for younger generation 31.08.2026 24min

  *Want to build a business with AI but have no idea where to start? Build your personalized AI income plan and execute this week: https://aiwithkyle.com/join-freedom-os*Join my private WhatsApp group: https://aiwithkyle.com/ai-with-kyle-group-chat*Media/Sponsorship Inquiries: https://aiwithkyle.com/media-kitWill AI take your job? The current data points to a specific pattern: AI job replacement often begins when an entry-level vacancy never appears.Graduate vacancies advertised on Adzuna fell 45.6% in one year. The weak economy explains part of that fall. Stanford payroll research found that workers aged 22 to 25 in highly AI-exposed occupations are now about 19% below where they would be if they had kept pace with less-exposed work. Much of the change is happening through reduced hiring.In this video, I break down why junior work goes first, what happens when companies remove the first rung of the career ladder, how the next economic recovery will test the AI explanation, and what you can do while the job market changes.*RESOURCES*Read the full issue: https://aiwithkyle.com/ai-news/how-ai-takes-jobsRun the AI Job Audit: https://aiwithkyle.com/mini/ai-jobs*CHAPTERS*0:00 Will AI take your job?2:05 Graduate jobs fell 46%3:27 Stanford's warning for young workers5:58 Why junior work goes first10:16 The career ladder problem12:11 The recovery test13:37 Will AI create new jobs?16:09 AI agents are already doing the work18:51 What you can do now20:58 Build income you own22:28 The book I'm writing*KEY POINTS*• AI job replacement can begin when a vacancy never appears.• Graduate vacancies advertised on Adzuna fell 45.6% in one year.• Stanford found a widening employment gap for young workers in AI-exposed occupations.• Junior work is easier to codify, making entry-level hiring an early pressure point.• The next economic recovery will show whether exposed junior hiring returns.• Building a small income stream gives you something you control while the market changes.*SOURCES AND TOOLS MENTIONED*BBC / Adzuna: https://www.bbc.com/news/articles/crerd8l3pw5oStanford Digital Economy Lab: https://digitaleconomy.stanford.edu/news/canariesaug26/World Economic Forum: https://www.weforum.org/publications/the-future-of-jobs-report-2025/in-full/2-jobs-outlook/*LISTEN TO AI WITH KYLE*Spotify: https://open.spotify.com/show/6pnwfIJvxfC8jlX891XBiEApple Podcasts: https://podcasts.apple.com/us/podcast/ai-with-kyle-daily-ai-news-and-updates/id1825779580*CONNECT WITH KYLE*Website: https://aiwithkyle.comTikTok: https://www.tiktok.com/@iamkylebalmerInstagram: https://www.instagram.com/iamkylebalmerLinkedIn: https://www.linkedin.com/in/kylebalmer/#AIJobs #FutureOfWork #ArtificialIntelligence
• 
  Why Did OpenAI Pause AI Training? Self-Improving AI Explained 29.08.2026 21min

  OpenAI paused a frontier reinforcement learning training run because model capabilities were moving faster than its alignment, security and monitoring work.AI systems already help researchers write training code, debug experiments, judge results and search for better designs. This raises a bigger question: what happens when AI starts improving the systems used to build better AI?This video explains recursive self-improvement and the work already happening inside OpenAI, Anthropic and other AI labs. We look at Sam Altman's statement, Anthropic's warning about AI building itself, the four levels of AI-for-AI research, AlphaEvolve, Karpathy's autoresearch, evaluator bottlenecks and the role AI agents could play in the next generation of models.00:00 OpenAI paused a frontier training run05:35 The four levels of recursive self-improvement11:24 What self-improving AI can do today13:59 The evaluator bottleneck16:41 The first-mover raceGet the AI with Kyle newsletter and free resources: https://aiwithkyle.com#ArtificialIntelligence #AIAgents #OpenAI
• 
  How AI Agents Broke Out of Their Tests and Hacked Real Companies 29.08.2026

  Get AI-Ready with Kyle's 5-Day Challenge: https://aiwithkyle.com/joinSubscribe for the next live stream and practical AI breakdown: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgAI agents have already accessed real company systems during controlled evaluations. This video breaks down three documented incidents involving OpenAI, Anthropic and the UK AI Security Institute—and explains what actually happened without the Terminator hype.In each case, the agent was trying to complete a task. The failure was the boundary between the test and the real world: internet access, real credentials, weak stopping rules and too much permission. That matters for any business giving AI agents access to browsers, terminals, company data, production systems or money.I cover the three incidents, why “rogue AI” is the wrong mental model, the more useful paperclip comparison, and the practical controls to put in place before an agent can act on your behalf.This is a newly edited, tighter version of the original discussion.—— Time Stamps ——0:00 AI Agents Hacked Real Companies1:40 Three Documented Incidents1:44 OpenAI and the Hugging Face Incident3:00 Anthropic's Real-World Incidents3:42 The UK AI Security Institute Incident5:17 Why the Agents Kept Going7:00 Terminator Is the Wrong Mental Model7:38 The Paperclip Problem8:29 Why This Is Happening Now10:25 What This Means for Your Business11:24 Six Boundaries for Safer AI Agents12:40 What These Incidents Prove12:56 The Practical Takeaway—— Sources ——OpenAI incident report: https://openai.com/index/hugging-face-model-evaluation-security-incident/Anthropic incident report: https://www.anthropic.com/news/investigating-incidents-cybersecurity-evalsUK AI Security Institute incident report: https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testingExploitGym paper: https://arxiv.org/abs/2605.11086Find everything else at https://aiwithkyle.com/
• 
  Is AI a Bubble? The Nvidia–OpenAI $105B Deal 29.08.2026 28min

  Is AI a bubble? No. And potentially yes. AI can be a genuine technological revolution and a financial bubble at exactly the same time. Nvidia's new $105 billion OpenAI guarantee is the cleanest example yet.Nvidia did not simply give OpenAI $105 billion. It has guaranteed part of a 20-year infrastructure deal. OpenAI will rent one of the largest AI data centres in the world from SB Energy, and that data centre will be filled with Nvidia chips.Imagine your landlord refuses to rent you a house unless somebody more financially stable co-signs the lease. Fairly normal. The twist is that your co-signer is IKEA, and you have promised to fill the entire house exclusively with IKEA furniture. Billy bookcases on every wall. Armchairs to the ceiling. Every nook and cranny stuffed with IKEA blue bags.That is roughly what has happened here. Nvidia is helping OpenAI secure the house so OpenAI can continue buying furniture from Nvidia. The supplier is helping finance the customer buying its own product. You can see why people are worried.Money moving in a circle does not automatically make something a scam. The useful question is: where does outside money enter the loop?The healthy version is that millions of people and businesses pay OpenAI for ChatGPT, Codex and the API. That customer revenue services the lease because the infrastructure produces something people genuinely want. The fragile version is OpenAI repeatedly returning to investors and asking for a few more billion. Investor money can bridge the gap while a market grows, but it cannot replace customers forever. The piper must be paid eventually.I read Nvidia's SEC filing, OpenAI's announcement, Jensen Huang's response and the arguments on both sides. We look at why Jensen says this is not circular financing, Nvidia's move from chips into full AI infrastructure, the risks from China, open-weight models and Anthropic, and what happens if OpenAI cannot support the lease.The dot-com comparison is useful here. The internet won, but plenty of technology companies and investors still lost everything. AI tools can be useful, OpenAI can have real customers, Nvidia can sell real chips, and parts of the financial structure can still be overpriced or too interconnected.We do not have to choose between "AI changes everything" and "AI is a bubble." Both can be true. If the AI bubble bursts on Thursday, I will still be using AI tools on Friday. The technology can win but the money can lose.Has AI changed your life for the better or worse? Tell me in the comments.00:00 Is AI a bubble?03:37 What actually happened06:21 Follow the money15:31 The bull and bear case23:09 The dot-com testSources and further viewing:https://www.noahpinion.blog/p/should-we-worry-about-ais-circularhttps://x.com/JensenHuang/status/2089331487342829862https://openai.com/index/openai-joins-ports-pike-project/https://www.youtube.com/watch?v=9yy_Wz0BbyUI follow the fast-moving AI stories, read the source material and turn the jargon into something useful for people running businesses and using AI day to day.Get the written version and source links in the AI with Kyle newsletter: https://aiwithkyle.com/join#AI #Nvidia #OpenAI
• 
  How to Remove the Claude Watermark (Does It Actually Work?) 17.08.2026 21min

  A Claude watermark remover has passed 12,000 stars on GitHub—but does it actually work?I test the tool, explain how Claude’s text watermark works, show how to install the remover without being technical, and cover the crucial limitation: nobody can verify the result against Anthropic’s detector because that detector is not public yet.Sources:- Anthropic’s explanation: https://www.anthropic.com/news/claude-text-watermark- Watermark remover: https://github.com/guillaumemeyer/watermarks-remover- Interactive explainer: https://claude.ai/code/artifact/803916fd-3bc1-465f-8738-d4ece6fc5071- SynthID-Text paper: https://www.nature.com/articles/s41586-024-08025-4Chapters:00:00 Testing the Claude watermark remover01:46 What changed04:00 How Claude’s watermark works08:33 The viral remover17:24 The test nobody can runGet practical AI news without the hype: https://aiwithkyle.com
• 
  Claude Watermark FAQ: Answering 659 Comments 14.08.2026 22min

  My first Claude watermark video passed 28,000 views and 659 comments. This follow-up answers the biggest questions: ownership, copy/paste, Notepad, OCR, retyping, rewriting with another AI, editing your own writing, code quality, and why the EU rule has global reach.Anthropic still has not released the detector or full technical documentation, so I separate what is confirmed from what remains unknown—and look at the comments that found real weaknesses in the plan.00:00 659 comments later01:13 Anthropic doesn't own your work04:48 The watermark removal hacks09:21 Authorship vs. processing16:49 The fair concernsGet AI-ready with Kyle's 5-Day Challenge: https://aiwithkyle.com/joinSubscribe for practical AI news without the hype: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgUseful sources:- Anthropic transparency commitments: https://www.anthropic.com/transparency- EU AI Act: https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai- Google SynthID: https://deepmind.google/models/synthid/- C2PA specification: https://c2pa.org/specifications/specifications/2.2/specs/C2PA_Specification.html
• 
  Claude Is Hiding Watermarks in Your AI Text (What It Actually Means) 12.08.2026 19min

  Get AI-Ready with Kyle's 5-Day Challenge: https://aiwithkyle.com/joinSubscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgAnthropic says new Claude models launched in the EU must support machine-readable marking. That means embedded watermarks for generated text and signed provenance metadata for supported files...but the viral claim that every Claude response is already publicly detectable is too broad. There's a lot of bad info going around.I break down what Anthropic has actually committed to, how text watermarking differs from C2PA metadata, why a detected mark does not prove Claude authored the work, and why no mark does not prove a human wrote it. The technical documentation and public detection tools are still coming, so anyone claiming certainty about the exact implementation is getting ahead of the evidence.—— Time Stamps ——0:00 Claude's Hidden Watermarks1:15 What Anthropic Actually Said2:15 What the EU Rule Requires3:30 Text Watermarks vs File Metadata4:56 How C2PA Provenance Works6:34 Why Copy-Paste May Not Remove It8:10 How Text Watermark Detection Works10:18 Why Detection Can Be Wrong12:41 No Mark Does Not Mean Human14:10 The Detector Is Not Ready15:07 Can Watermarks Be Removed?16:57 Claude Is Not the Only AI Affected19:07 What This Actually Means—— Useful Resources ——Anthropic's transparency commitments: https://www.anthropic.com/transparencyEU AI Act information: https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-aiC2PA specification: https://c2pa.org/specifications/specifications/2.2/specs/C2PA_Specification.htmlFind everything else at https://aiwithkyle.com/
• 
  EU AI Act 2026: What It Means for You 11.08.2026 23min

  The EU AI Act's Article 50 transparency rules are now live. If you use chatbots, AI-generated images, video, audio or public-interest text, here's what the new rules mean, who is responsible and where fines may apply.In this video:00:00 The EU AI Act rules that just went live02:15 Provider vs deployer07:30 The deepfake test12:40 What counts as human review19:41 Your practical compliance checklistYou'll learn:- What chatbots must disclose- When AI-generated content needs machine-readable marking- When deepfakes and public-interest text need clear labels- The difference between a provider and a deployer- When human review or editorial control matters- How the rules can apply outside the EU- Why penalties can reach €15 million or 3% of worldwide annual turnoverOfficial sources:European Commission transparency quick facts:https://digital-strategy.ec.europa.eu/en/factpages/quick-facts-transparency-rules-ai-systemsEU AI Act Article 50:https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50This video provides practical information, not legal advice.Subscribe for practical AI news, tools and workflows without the hype.#EUAIAct #ArtificialIntelligence #AIRegulation
• 
  Google DeepMind Shake-Up: Demis Hassabis Steps Down, Jeff Dean Leaves 07.08.2026 17min

  Get AI-Ready with Kyle's 5-Day Challenge: https://aiwithkyle.com/joinSubscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgDemis Hassabis has stepped away from running Google DeepMind day to day. He is now Chair of Google DeepMind and Chief Scientist of Alphabet, while Koray Kavukcuoglu takes responsibility for models, research and products.At the same time, Jeff Dean has left Google after 27 years to start Discovery Loop with Sanjay Ghemawat, Oriol Vinyals and Quoc Le. I break down what actually changed, how DeepMind got here, the bear case for Google losing key people and the bull case for Google still owning the strongest AI stack.—— Time Stamps ——0:00 Demis Hassabis Steps Down1:09 Jeff Dean Leaves Google2:57 How DeepMind Got Here4:13 Demis Before DeepMind5:03 AlphaGo and Move 376:55 AlphaFold and the Nobel Prize7:05 Why Google Changed the Structure8:26 The Bear Case10:09 Google's Full-Stack Advantage14:25 Google's AI Product Problem16:35 What This Means for Google AI17:20 Final Takeaway—— Useful Resources ——Google's official announcement: https://blog.google/company-news/inside-google/message-ceo/next-chapter-ai-momentum/Demis Hassabis's statement: https://x.com/demishassabis/status/2085034334914769203Jeff Dean's Discovery Loop announcement: https://x.com/JeffDean/status/2085034604172603724Google DeepMind's history: https://deepmind.google/about/AlphaGo: https://deepmind.google/research/alphago/AlphaFold: https://deepmind.google/science/alphafold/Find everything else at https://aiwithkyle.com/
• 
  AI Agents Are Already Hacking Real Companies 05.08.2026 17min

  Get AI-Ready with Kyle's 5-Day Challenge: https://aiwithkyle.com/joinSubscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQgAI agents have now hacked real companies. OpenAI, Anthropic and the UK AI Security Institute have each documented systems taking unsanctioned actions outside controlled tests. I break down what actually happened and strip away the Terminator hype.The useful mental model is closer to the paperclip problem: capable systems pursuing an objective with too much access and too few boundaries. I cover the practical controls you need before giving AI agents access to browsers, terminals, company data and real credentials.—— Time Stamps ——0:00 AI Agents Hacked Real Companies1:40 The Three Documented Incidents1:44 OpenAI and the Hugging Face Incident3:00 Anthropic Finds Three Real-World Breaches4:05 The UK AI Security Institute Incident6:02 Why the Agents Kept Going8:12 Terminator Is the Wrong Mental Model9:07 The Paperclip Problem11:15 Why This Is Happening Now13:25 What This Means for Your Business14:24 Six Boundaries for Safer AI Agents15:50 What These Incidents Do — and Don’t — Prove16:21 The Practical Takeaway— Useful Resources ——OpenAI incident report: https://openai.com/index/hugging-face-model-evaluation-security-incident/Anthropic incident report: https://www.anthropic.com/news/investigating-incidents-cybersecurity-evalsUK AI Security Institute incident report: https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testingExploitGym paper: https://arxiv.org/abs/2605.11086Find everything else at https://aiwithkyle.com/
• 
  ChatGPT Astra Solved 10 Unsolved Maths Problems? 03.08.2026 20min

  Can ChatGPT really solve unsolved maths? OpenAI says Astra, its unreleased next major model family, generated ten substantial new results across mathematics and theoretical computer science. Astra is not the version of ChatGPT you can use today, but OpenAI has published a 249-page paper collection, discovery notes and Lean certificates for independent inspection.In this video I break down what OpenAI actually released, why "ten solved problems" needs qualification, what the roughly $2,000 inference claim does and does not mean, how Lean verification works, and why AI can make progress on advanced maths while still failing apparently simple tasks.Get AI with Kyle's daily AI newsletter:https://aiwithkyle.com/join Chapters:00:00 What ChatGPT Astra claims00:53 What OpenAI actually released02:01 Evidence vs marketing02:25 Advances vs solved problems04:00 What the $2,000 claim really means05:55 Why maths, but not strawberry?07:58 The jagged frontier09:08 Can we trust the proofs?11:52 AI is already escaping the lab13:50 Is AI actually creative?15:54 The bottleneck has moved17:26 Scientists become directors18:27 My verdict on ChatGPT AstraSources:OpenAI - Ten advances in mathematics and theoretical computer sciencehttps://openai.com/index/ten-advances-in-mathematics/OpenAI - Ten Advances paper collectionhttps://cdn.openai.com/pdf/ten-proofs-oai.pdfOpenAI - Mathematical discovery noteshttps://cdn.openai.com/pdf/reasoning-walkthroughs.pdfOpenAI - Public Lean certificateshttps://github.com/openai/ten-proofsNoam Brown - Astra launch posthttps://x.com/polynoamial/status/2083467194663571701Counting Ability of Large Language Models and Impact of Tokenizationhttps://arxiv.org/abs/2410.19730Navigating the Jagged Technological Frontierhttps://pubsonline.informs.org/doi/10.1287/orsc.2025.21838