Latest / AI Fire Daily / 🎙️ EP 282: Microsoft Clones OpenClaw with "Scout" & Bernie Sanders Wants 50% of OpenAI
Transcript
- 0:00Imagine having a permanent AI coworker who knows
- 0:03your unique quirks, it never sleeps, and it manages
- 0:07your daily workflow. Right. Just an always -on
- 0:09digital colleague. Exactly. But now imagine the
- 0:12federal government suddenly taking a 50 % ownership
- 0:15stake in it. Welcome to the Deep Dive. Yeah,
- 0:18we are cutting through a lot of AI industry noise
- 0:20today. We want to give you the exact insights
- 0:22you actually need. Today, we are exploring Microsoft's
- 0:26massive new AI agent called Scout. Then we will
- 0:29unpack the rapid physical expansion of global
- 0:32AI infrastructure. It is a massive shift. It
- 0:35really is. And finally, we will examine a groundbreaking
- 0:37political proposal by Bernie Sanders regarding
- 0:40sovereign wealth. The pace of these industry
- 0:42advancements is genuinely staggering right now.
- 0:45We are moving from digital software into physical
- 0:48infrastructure so fast. We have an incredible
- 0:50amount of dense material to cover today, but
- 0:52we are going to break it all down. down for you.
- 0:54Let us begin with the micro level of your daily
- 0:56life. Yeah, Microsoft recently dropped something
- 0:58massive at their Build Developer Conference.
- 1:01They introduced a permanent digital colleague
- 1:04that they officially named Scout. Right. It is
- 1:07an always -on personal assistant. And it lives
- 1:09entirely inside the Microsoft 365 ecosystem.
- 1:13Microsoft was clearly taking very detailed notes
- 1:16on the OpenClaw framework. You might remember
- 1:18the absolute chaos of OpenClaw earlier this year.
- 1:21Oh, absolutely. The chaos and joy of that whole
- 1:23release. Exactly. Developers were thrilled, but
- 1:26the autonomous agents were just highly unpredictable.
- 1:29Microsoft wanted that immense autonomous power,
- 1:32but with much better corporate control. So Scout
- 1:35provides you with a truly persistent digital
- 1:37identity every day. It maintains a consistent
- 1:40working style across all your software applications.
- 1:42Yeah, and you even give your specific instance
- 1:44its own customized human name. You might decide
- 1:47to name your personal digital assistant Sebastian
- 1:49today. And then you train Sebastian over time
- 1:52through your daily feedback loop. It learns exactly
- 1:55how you prefer to format your weekly reports.
- 1:57It is kind of like stacking Lego blocks of data.
- 2:00Oh, I like that analogy. Yeah, to build the perfect
- 2:03digital colleague, you actively correct them
- 2:05when they format a financial report incorrectly.
- 2:08Over several weeks, they adapt to your highly
- 2:10specific micro preferences perfectly. The deep
- 2:13workflow integration here is, honestly, quite
- 2:16a massive leap. It lives entirely in the cloud,
- 2:18but it runs across your desktop. It seamlessly
- 2:21hooks into your web browser and your daily inbox.
- 2:24But getting access to Scout does require jumping
- 2:26through some early hoops. You must be accepted
- 2:29into Microsoft's exclusive Frontier program for
- 2:32early adopters. Right. It also requires an active
- 2:35GitHub co -pilot subscription to function properly.
- 2:38They are tightly controlling the initial rollout
- 2:40of this powerful agentic technology. And for
- 2:43good reason, honestly. We all know that autonomous
- 2:46AI agents can run completely amok. I still wrestle
- 2:49with prompt drift myself. Oh, totally. It happens
- 2:51to all of us. You ask an AI to do one thing and
- 2:54it just wanders. Anyone remember the OpenClaw
- 2:57inbox horror stories from a few months back?
- 2:59Yeah, where agents were accidentally deleting
- 3:01massive email threads. Yeah. Completely wiping
- 3:04them out without any human oversight. Microsoft
- 3:06explicitly wanted to avoid those exact corporate
- 3:09enterprise horror stories. To prevent that, they
- 3:12built a strict policy conformance system. Right.
- 3:15This system acts as a hard band relayer for the
- 3:18software. It forces the AI to request permission
- 3:21before finalizing destructive actions. But wait,
- 3:24an automatic auditor sounds great in theory for
- 3:27companies. If Sebastian is already deleting files,
- 3:30a paper trail just places blame. How does a conformance
- 3:33system actually stop unpredictable AI behavior
- 3:36beforehand? Well, it actively intercepts the
- 3:38underlying software commands before they ever
- 3:40execute. It analyzes the intended action against
- 3:43a strict set of corporate rules. If the action
- 3:45violates a rule, the execution is instantly blocked.
- 3:48So it's an automatic auditor leaving a strict
- 3:50paper trail. Exactly. It provides actual hard
- 3:53limits for cautious corporate IT departments.
- 3:56The paper trail is just the secondary logging
- 3:58mechanism for IT. If Scout is handling our desktop
- 4:01workflows, that requires immense background compute.
- 4:05That brings us to the physical reality of all
- 4:07this digital convenience. Yeah, AI is rapidly
- 4:10expanding its footprint across global physical
- 4:12infrastructure today. OpenAI just launched six
- 4:15brand new codex tools for the workplace. And
- 4:18these are built entirely for white -collar work
- 4:20without any coding required. They specifically
- 4:22target sales, finance, design, complex analytics,
- 4:26and creative roles. You just talk to the system,
- 4:28and it generates the work. workflow. Meanwhile,
- 4:31The Verge has been actively testing Gemini Spark
- 4:33from Google. They are calling it the most terrifying
- 4:36and impressive AI experience. They say it is
- 4:39the closest thing to a truly useful agent. Yeah,
- 4:42it can navigate complex websites and execute
- 4:44multi -step tasks completely autonomously. Google
- 4:47also gave their Gemma agents something surprisingly
- 4:50useful this week. Right. They introduced reusable
- 4:52digital skills that agents can load on demand.
- 4:55This specific code repository makes lightweight
- 4:57AI agents significantly smarter overall. They
- 5:00do not need to hold every single skill in active
- 5:02memory. They simply fetch the specific digital
- 5:04tool required for the current task. The open
- 5:07source community is also moving incredibly fast
- 5:09right now. Minimax just released M3, which is
- 5:12their latest open -weight AI model. Instead of
- 5:15hoarding the underlying recipe like the massive
- 5:17corporate labs do, they release the actual mathematical
- 5:20connections the AI learned during training. This
- 5:23allows developers to fine -tune the model locally
- 5:25without needing supercomputers. Minimax claims
- 5:28it completely outperforms GPT -5 .5 and Gemini
- 5:313 .1 Pro today. Yeah, it is dominating on several
- 5:35rigorous coding benchmarks across the entire
- 5:37industry. It is rapidly closing in on the performance
- 5:40of Cloud Opus 4 .7. But this rapid software expansion
- 5:43is bringing significant new security risks. That
- 5:47intense desire to download models locally creates
- 5:50a massive vulnerability. Security researchers
- 5:53just found multiple fake Cloud code installer
- 5:56sites online. These malicious sites are specifically
- 5:59designed to steal your sensitive credentials.
- 6:02They aggressively target your secure password
- 6:04vaults and your development environments. They
- 6:06are specifically hunting for your highly valuable
- 6:08development API keys. And you really have to
- 6:11protect those. Definitely. API keys are secret
- 6:13digital passwords that let different software
- 6:15programs talk safely. You must double -check
- 6:18every single download link before installing
- 6:20new tools. To avoid relying on local downloads,
- 6:23you need massive cloud compute power. Anthropic
- 6:26just expanded their cloud mythos system to 150
- 6:29organizations. We are not talking about software
- 6:32startups running basic customer service chatbots.
- 6:34We are talking about critical infrastructure
- 6:36systems across 15 different countries. These
- 6:40organizations run national power grids, global
- 6:42telecom networks, and security. Integrating advanced
- 6:46autonomous AI into a national power grid is a
- 6:49massive shift. To support all this incredible
- 6:51computing power, OpenAI is breaking ground quickly.
- 6:55They are building a massive Stargate data center
- 6:57campus in Michigan. They are calling this massive
- 6:59new computing facility the barn. Whoa. Yeah.
- 7:03Imagine scaling a one gigawatt data center just
- 7:05for AI. Two sec silence. It is just a staggering
- 7:08amount of energy. A one gigawatt facility requires
- 7:11an almost unfathomable amount of physical electricity.
- 7:14That is enough power to run a medium -sized American
- 7:17city comfortably. It is expected to create 2
- 7:19,500 local tech jobs. OpenAI is also offering
- 7:23$45 million in educational codex credits. Those
- 7:27computing credits will go directly to local students
- 7:29in the area. The scale of physical investment
- 7:31here is completely unprecedented in tech history.
- 7:35You have these gigawatt facilities being built.
- 7:37Right, and it perfectly highlights the incredible
- 7:40tension in the industry today. You have massive
- 7:43gigawatt data centers being built alongside serious
- 7:46security vulnerabilities. The Michigan Stargate
- 7:48campus is huge, but then you have these fake
- 7:51installer risks. It is a wild juxtaposition.
- 7:54Because AI is affecting the real world, the government
- 7:57is stepping in. After largely staying hands -off,
- 8:00President Trump just signed a new executive order.
- 8:02This sweeping order seeks comprehensive federal
- 8:05oversight of these massive AI models. It creates
- 8:08a formal regulatory framework for deploying systems
- 8:10affecting critical infrastructure. What is truly
- 8:13fascinating here is the unified response from
- 8:15the tech industry. OpenAI, Google, Microsoft,
- 8:18and Anthropic all publicly supported this new
- 8:21executive order. Yeah, they are openly welcoming
- 8:23this formal regulatory framework from the federal
- 8:25government. It indicates a transition from unregulated
- 8:28growth to structured corporate compliance. We
- 8:31are also seeing a massive push into real -world
- 8:35physical robotics everywhere. A robotics company
- 8:38called Mecca AI just raised $60 million. They're
- 8:42trying to help physical robots learn complex
- 8:45human actions. The company collects precise movement
- 8:48data directly from wearable body sensors. They
- 8:51also use standard iPhones to capture detailed
- 8:54real -world physical interactions. They use this
- 8:57massive data set to train robots for manual physical
- 9:00tasks. The sheer scale of this technology is
- 9:02simply outpacing the existing frameworks. The
- 9:05competition between open -weight models and closed
- 9:07proprietary systems is genuinely heating up.
- 9:10How are smaller models like MINIMAX M3 competing
- 9:12with these massive gigawatt giants? Well, they
- 9:15utilize highly optimized training techniques
- 9:17rather than relying Welcome back to our deep
- 9:35dive into the shifting AI landscape. We just
- 9:38discussed companies building massive gigawatt
- 9:40scale data centers in Michigan. Yeah, and we
- 9:43explored permanent AI coworkers that can easily
- 9:46replace routine human tasks. The financial valuation
- 9:50of these top AI labs is absolutely exploding
- 9:53right now. This incredible financial boom leads
- 9:55directly to the ultimate trillion dollar question
- 9:58today. Who actually owns all of this newly created
- 10:02economic value globally? Bernie Sanders just
- 10:05dropped a fascinating op -ed in The New York
- 10:07Times. He is previewing a wildly ambitious new
- 10:11legislative bill for the public. It is officially
- 10:13called the American AI Sovereign Wealth Fund
- 10:16Act. Instead of letting a few tech billionaires
- 10:19capture all the upside, he pivots. Right. He
- 10:21wants the federal government to treat artificial
- 10:23intelligence like a national resource. If you
- 10:26have been tracking the massive valuations of
- 10:28open AI, you know. The average retail investor
- 10:30is completely locked out of these massive gains.
- 10:32You cannot easily buy shares in these highly
- 10:34exclusive private AI companies. Sanders is arguing
- 10:37a very specific philosophical point regarding
- 10:40the training data. He notes that these powerful
- 10:43models were trained on collective human knowledge.
- 10:46They scrape the public Internet, which represents
- 10:48the accumulated labor of humanity. Therefore
- 10:51Sanders argues that humanity fundamentally deserves
- 10:55a financial cut of the profits. It is a really
- 10:58interesting perspective. We are treating abstract
- 11:00human knowledge much like we treat physical timber.
- 11:04You find the resource you harvest it and you
- 11:07sell it for profit. The proposed mechanics of
- 11:09this sovereign wealth bill are actually. Quite
- 11:12fascinating. It would collect a massive one -time
- 11:14tax from these top AI labs. However, it would
- 11:18not collect this tax in the form of traditional
- 11:20cash. The bill demands the corporate tax be paid
- 11:23in the form of company stock. Yeah, which would
- 11:26immediately give the federal government massive
- 11:28corporate voting power. They would literally
- 11:30secure a powerful board seat at the top AI labs.
- 11:34That is not just about money. It is about active
- 11:36corporate governance. If the government has a
- 11:38board seat, they vote on safety releases directly.
- 11:41Sanders points to some very successful historical
- 11:43precedents to justify this ambitious move. He
- 11:46points to Norway's massive $2 trillion sovereign
- 11:50public oil fund for context. They found massive
- 11:53oil reserves, taxed the extraction, and invested
- 11:56it for citizens. Exactly. He also references
- 11:59Alaska's longstanding... permanent oil dividend
- 12:02program for its residents. I hear Sanders' argument
- 12:05about our data training these massive models.
- 12:08But a tech founder would argue they built the
- 12:11multi -billion dollar processing infrastructure.
- 12:13That is a very fair counterpoint. If I build
- 12:16a public library, do I owe the paper manufacturer
- 12:18royalties? Do I owe the person who planted the
- 12:21original tree for the paper? That is exactly
- 12:24the argument the massive tech companies will
- 12:25likely use. But there is a brilliant piece of
- 12:28political irony embedded in Sanders' argument.
- 12:31part about universal income. Yeah. Sanders hilariously
- 12:34points out something the prominent AI founders
- 12:36do constantly online. These tech founders love
- 12:39to hype up universal high income in their manifestos.
- 12:41They frequently promise that AI will eventually
- 12:44generate massive wealth for everyone. They claim
- 12:47this technology will eventually usher in a post
- 12:49-scarcity utopian future. Sanders is essentially
- 12:52calling their massive bluff on these grand utopian
- 12:56promises. He is saying, if you truly believe
- 12:59that, put your equity up now. Public sentiment
- 13:02around artificial intelligence is pretty rocky
- 13:05across the country right now. The deep public
- 13:07anxiety about imminent job displacement is incredibly
- 13:11real for workers. So the core philosophical argument
- 13:14actually makes a lot of sense to average people.
- 13:16If our collective Internet data trained these
- 13:19models, we should share the wealth. It is a brilliant
- 13:22political chess move that traps the industry
- 13:24completely. They cannot easily dismiss this without
- 13:27looking incredibly greedy to the general public.
- 13:29It forces the entire tech industry. to answer
- 13:31a very tough philosophical question. Is this
- 13:34revolutionary technology truly for everyone or
- 13:37just the elite corporate shareholders? It is
- 13:39a policy proposal directly addressing deep public
- 13:42anxiety about job displacement. What makes this
- 13:44a particularly difficult political trap for AI
- 13:47founders to escape? Well, if they fight the proposed
- 13:50bill. They directly contradict their own published
- 13:54utopian visions. They cannot promise a post -scarcity
- 13:58future while simultaneously hoarding all the
- 14:00private equity. It uses the founders' own utopian
- 14:03promises of universal income against them. Beat.
- 14:07It absolutely does. It weaponizes their own optimistic
- 14:10tech manifestos against their current financial
- 14:13interests perfectly. We have covered a massive
- 14:17amount of highly complex ground here today. We
- 14:19started by examining how to train a personalized
- 14:21scout on your desktop. We explored how it builds
- 14:24a permanent digital identity for your daily workflow.
- 14:27Then we zoomed out to look at the gigawatt scale
- 14:29infrastructure powering it all. We examined the
- 14:32aggressive expansion of open -weight models and
- 14:34physical robotic training systems. We also highlighted
- 14:37the growing security risks surrounding fake download
- 14:40sites and API keys. We looked at how models are
- 14:43now running national power grids and telecom
- 14:45networks. Finally, we explored the deep philosophical
- 14:47and legal battle brewing in Washington. We examined
- 14:51Bernie Sanders' proposal regarding who actually
- 14:53owns the mind of the machine. The landscape of
- 14:55artificial intelligence is fundamentally shifting
- 14:57under our feet every single day. It really is.
- 15:00Before we wrap up today, I want to leave you
- 15:02with one final provocative thought. We discussed
- 15:05how our collective human knowledge trained these
- 15:07trillion -dollar AI models already. If that is
- 15:12true, how should you value your own daily digital
- 15:16footprint? Every single click, post, and keystroke
- 15:20is personal data fueling this massive engine.
- 15:24How will you protect and value your personal
- 15:26data going forward in this new era? That is a
- 15:29profoundly important question to carry with you
- 15:31throughout the week. Thank you for joining us
- 15:33for this deep dive into the AI landscape. We
- 15:35hope we helped you cut through the noise and
- 15:37find clarity today. Keep questioning the massive
- 15:39digital systems being built all around you right
- 15:41now. Stay curious, protect your data, and we
- 15:45will see you next time. Utero Music.