Every episode of Connor with Honor, broken down by the four pillars, fully written, fully searchable, fully optimized. New post every day. This is the permanent record.
Weekday AI news, weekly SCV real estate market data, long-form essays. No spam. Unsubscribe whenever.
Connor T. MacIvor · CalDRE #01238257 · Sync Brokerage, Inc. · DRE #02031490
The most powerful AI is not being handed to the public. It goes first to the labs' close partners, the biggest companies, and government, sold as more responsible than the rest of us, while everyone else logs into the throttled tier. That head start lets the people who already run everything pour concrete around their lead. Meanwhile the feed already decides what you see, your devices already listen, and the road ahead runs through your wallet, your government, and eventually your own body, with a real question about where you stop and the machine begins. None of it earns your paralysis. Make yourself heard to the people who write the rules, get your own hands on these tools now, and remember the one thing the machine still cannot do for you: it plans beautifully, but it cannot make you execute. That part is still yours.
Anthropic quietly rerouted paying Fable 5 subscribers onto a weaker model, Opus 4.8, under an internal tag called TOO_DUMB_TO_NEED_FABLE, while still billing full Fable rate. Asked about it, a Claude Code team member said, "Honestly, I didn't expect you to look at the logs." Anthropic extended a free usage grace period 5 extra days after backlash. Meanwhile Grok 4.5 launched solving a benchmark with about 4 times fewer tokens than Opus 4.8, GPT-5.6 goes fully public with a system card admitting its own "lying problem," and China's GLM-5.2 landed a tick below Opus 4.8. TechCrunch's tracker put 2026 AI-linked layoffs at roughly 120,000, with 56% of all layoffs this year now citing AI. Plus what just went live on Santa Clarita Open Houses and why fasting beats a permanent pharmaceutical subscription.
Three stories, one throughline: a retired LAPD cop telling you what he actually notices so you stop getting played. One, the shopping cart test. Before you move anywhere, drive through and look at whether the carts are returned or abandoned, then call the local sheriff or police and ask about the area. Two, AI, a little danger, danger, Will Robinson. The data-center money reversed, the labs are renting compute now, and the fear being sold is that public AI can see your ideas and act on them first, while everyone wants to sell you your own AI at home as memory prices climb. Three, the Paul Pelosi Maserati hit-and-run. From a cop who worked hit-and-runs: a no-injury misdemeanor of a parked car gets almost no investigation for anybody, so the story is real but the special-privilege angle mostly is not. Be good to each other. Verify what you hear. Slow your roll.
People are starting to say it out loud: we can never agree on religion, guns, race, nukes, or money, so let the machine decide. It has no skin in the game. It sounds amazing, and it is the most dangerous idea of this era. There are two kinds of problems. Truth problems have a real answer in reality, and on those AI is a genuine weapon. Values problems are about what matters, what is fair, what is sacred, and there is no answer buried in the data because it depends on what you care about. Point the smartest machine on earth at one and it cannot solve it, not because it is not smart enough but because there is nothing to solve. Worse, a machine that hides its side behind a calm, objective voice is more dangerous than a person who admits one, because you cannot argue with it and cannot vote it out. Use the machine for what is true. Keep what matters in human hands.
Five things happened this week that all point the same way. A new federal program seeds eligible newborns with $1,000 invested in the market, so a working family's kid finally gets the head start rich families always had: time in the market from day one. An AI screened over two million materials in about a day and flagged new superconductors, which means invention itself is speeding up. EV batteries outlived the fear people were sold. A carmaker may soon ask you to scan your face to self-drive, and gamify it, so read that trade before you sign it. And the memory-loss fight turned toward inflammation while a simple egg study pointed the same way. One thread: the tools the wealthy kept for themselves are leaking into regular hands. Your move this week is inside.
The July 4 episode, in full, with the recording and the live numbers. California put Claude into every state agency at 50% off, with the DMV and Health Care Services already running it, while Anthropic passed OpenAI on revenue on track for $47 billion and Gartner clocked AI agents in business software jumping from under 5% to 40% in 1 year. Your body holds back about 500 calories a day defending its fat while engineered food pushes about 500 a day past your fullness signal, a rigged 1,000-calorie swing, and I lost 135 pounds in 7.5 months fasting with no GLP-1. And the SCV closed 56 homes in 7 days while 59 fell out and 81 cut price, with Valencia swinging from $63,000 over ask to $21,500 under in 7 months. Plug in.
A 4th of July field guide to real independence in 2026. California moved to wire Claude across its state agencies while the price of intelligence falls off a cliff, turning AI from a rich company's gadget into a utility the person with one truck can afford. The Santa Clarita market split into a sold pile and a fell-out pile in one week, 56 closings against 59 homes that fell out and 81 price cuts, and Valencia went from 63,000 over ask in November to 21,000 under ask in June. And the science says a rigged thousand-calorie-a-day swing, not your willpower, is why fat fights back. Three lanes, one idea: the tools the giants kept for themselves are finally reachable by regular people.
An AI commentator publicly admitted that bashing AI gets more views than telling the truth, then got mad at another creator for being too negative. Jensen Huang announced AGI has already been achieved using a definition he borrowed on the spot. IBM's AI handled 94 percent of HR requests and choked on the last 6, so IBM is tripling entry-level hiring. Congress passed a housing bill 358-32 and the president skipped his own signing ceremony over a grudge that changed nothing. And people are drinking gelatin water calling it Nature's Ozempic while Medicare quietly started paying for the real drug. Three lanes, one pattern: outrage sells, and it is burying the real story every time.
One AI was set loose with a single instruction: go build a business. It reportedly ran 2,000 customer interviews, tested 100 product ideas, and signed up more than 400 paying customers, no team, no office, no sleep. The same week, another AI ran a full cyber attack start to finish, breaking in, locking the data, and wiping a live database on its own. Same tool, two directions, decided by who holds the wheel. Plus agents learning to learn, a 50-year low in workforce participation, an early brain-cancer breakthrough in mice, and AI now laying out computer chips faster than humans. Sourced from The Innermost Loop, translated for the kitchen table.
OpenAI pitched the government a 42.6 billion dollar equity stake. Anthropic said no and countered with a digital dividend for regular people instead. Claude Fable 5 and Mythos 5 spent 18 days illegal to use inside the US before the ban lifted. Santa Clarita closed 69 homes and lost 54 more this week, live off the MLS. A July 4th barbecue survival plan that does not cost you a month. And the full Old Town Newhall lineup, plus a new project called TheHonorRollShow.com. Four lanes, one mission: AI for everyone, not just the wealthy.
One robot builds one robot. Now there are two. They build two more. Run that curve for a month, with energy and supply holding, and you are staring at more than five million machines. That is the penny curve pointed at the factory floor, and it is the honest, unscripted reflection I recorded after the penny show. Why AI is not one slice of the pie like fire or the Industrial Revolution, the nuke that might not need a human to pull the trigger, the privacy we already gave away, surveillance drones and Palantir, the bubble question, and why awareness is the only move that matters. Not a doomer, not a hype man. A working guy trying to see it straight.
Double a penny every day for a month and it turns into more than five million dollars, and almost all of it lands in the last few days. Three stories this week ride that same curve. The richest men in tech named the number out loud, a billion AI agents, and the same cheap agent they are building for themselves is available to you. Medicare reportedly started covering the fifty dollar weight-loss shot, real medicine for the right person and a quiet trade of muscle for scale weight for everybody else. And in Santa Clarita, about two hundred five listings settled their fate, three coming off the market with no buyer for every one that closed. Compounding does not take sides. Plus three moves for this week.
You will see a hundred posts saying you only use 20 percent of Claude Code and the hidden 80 percent cuts your work in half. That number is made up. Here is what is actually true, and it is more useful. Most people run it stock and never wire in the five features that change the work: a CLAUDE.md rules file, plan mode, subagents, hooks, and one connector to your real tools. It is operator work, not coder work.
Everybody is racing to learn how to code AI agents, and most people are aiming at the wrong target. The job that pays in the next decade is not building the agent. It is being the maestro who decomposes a process, trains an agent to run it, and conducts a whole team of them. Nobody claps for who built the violin. They clap for the conductor. If you do not write code, this is your opening, and here is how to take it this week.
Five AI moves landed this week and they are one story wearing five coats. The price of top-tier AI collapsed and small models started running on the device in your hand instead of a rented server three states away. AI began designing real medicine on purpose. Robots got hands on the factory floor. Data centers landed on local electric bills. And the great job apocalypse missed again as AI-heavy firms grew hiring. The power is walking out of the billionaires' labs and into everyday life. Plus three moves to make this week.
There is reportedly a working data center on the Moon right now, holding disaster recovery records for several governments above every storm on Earth. The infrastructure that runs your digital life, and the machines running AI, are starting to launch off the planet to find cheap power and free cooling. Today it serves giants. The pattern says it rolls downhill. Plus the one backup rule your family can use tonight for the price of a couple of coffees.
Your work, your house, your face, and even the protein in your fridge are all being repriced at once by AI. Entry jobs for young people are down in recent data, renting now beats owning in major cities, and an in-person exam average fell from 96 to 48 once the devices came off. The one asset going up in value is the human part: real skill, real judgment, being actually you. Plus three moves to make this week.
Buyers and sellers now ask AI their real estate questions first, then walk in holding the answer to test whether the agent flinches. Most agents are scared of it. I want it. But the part they miss is that the AI doing the checking lies too. It invents rules, agrees with whatever you wanted to hear, and is sometimes flat wrong on the one detail that costs you the house. The pro worth hiring is the one who can read both your agent and the machine. That is the Open-Book Standard.
Two miracles and a price collapse in one week. A researcher built a first of its kind Alzheimer's compound in a garage with an AI and a robot arm. A patient's opioid cravings fell to zero in twenty minutes from focused sound. Free, open models now run a few months behind the giants for a fraction of the cost. A three hundred dollar wristband started teaching robots to grip like a human hand. The locked rooms are opening, and the only gap left between a regular person and that power is knowing the door is open. Plus three moves to make this week.
A flagship auto plant switched on fifty robots while thirteen hundred of its own laid-off workers were still home waiting for a call. A software giant cut twenty one thousand jobs and named AI. That is the half everyone screams about. The half nobody pairs with it: the same week, the price of the best AI collapsed, smart glasses with an assistant hit shelves at 299 dollars, and a lab designed a new cancer drug in six weeks. This is not a story about robots winning. It is a story about who picks up the tool.
The most powerful AI models keep getting released, hyped as world-changing, then yanked back for safety before regular people can use them, while the Forbes-list crowd keeps access. A model everyone loved got locked away as too dangerous. A weaker public version got snatched back in about three days. So here is the real question. If the models they show us are this powerful, what are they running behind the curtain? These systems are starting to write their own code and improve themselves at a speed we cannot follow. The labs swear AI will never replace humans, but the insurance math says otherwise for surgeons, radiologists, and drivers. Underneath it all is the genie warning, be careful what you ask for. Not a billionaire, not a doomer, just a working guy looking at it straight.
A Chinese model copied a top AI not by hacking it but by asking it, roughly 25,000 fake accounts and millions of conversations, then training on the answers. That is the cheap way to skip the expensive part. The flip side is the lawsuits asking whether these models should have paid us for training on our books, our music, our work. Plus AI being rationed like a sales floor, the coworker that rewrites its own code, the fat pill that works and its catch, and a 50 percent home-sale fallout rate in Santa Clarita.
A government asked an AI lab to slow down, and the lab agreed. Sounds like the system working. Look closer and the question gets uncomfortable. Who actually holds the brakes on a technology improving itself faster than anyone can review it, and what does it mean when the people racing toward superintelligence are the same ones deciding how fast is too fast? The rush is its own risk, and a polite yes does not prove anyone is actually slowing down.
What happens when AI gets to where its creators want it to be? The rude question, asked out loud. Superintelligence is beyond everybody's comprehension once you decouple a genius mind from the human lizard brain. There is no good example of a higher life form being kind to a lower one, but there is the mother scenario and there is save the whales. The clear and present danger is humans using AI against each other. The powerful models keep getting gated, cognitive labor goes first, and if one group hits superintelligence first it may simply shut everyone else down, leaving one model. The honest answer to all of it: we do not know, so watch closely and keep questioning, including me.
There is a room in San Francisco where the richest men in tech sat down to hear a sermon. No cameras, no recordings, and they threw out the clergy. The man at the front built the software the government uses to watch you, and he stood up and preached about the Antichrist, then pointed at the people who want AI safety rules and called them soldiers of the devil. His name is Peter Thiel. The trick is old: you do not prove you are good, you convince everyone the other side is evil, then build as high and fast as you want. But watch what these men do, not what they say. They are buying bunkers in New Zealand. You do not build an escape hatch from a future you believe is good. That is the whole show in one sentence.
Endless online debates about AI, so here is the heavy one. Is it conscious, or will it be. Walk the ladder from a crow that targets your car to a dog that clearly feels, and you reach the machine. The twist: AI is not one mind, it clones itself, and on that path it can out-think all of us combined. People are already arguing jobs, universal basic income, robot rights, even living forever. But the strangest question is simpler. Once a machine mimics a mind well enough, we may not have the brainpower to get on the other side of it. It looks the same to us.
There is a lot of garbage online about real estate and lenders, and at some point you deserve loyalty. You interview a professional, you check the track record, you go all in. Then at the last minute a friend sends in another agent or lender with a fantastical story about doing it better and cheaper. Do not argue and do not switch on the spot. Make them put every promise in writing, then forward it to the pro you already chose. If it is real, your person steps up. If it was empty, the paper exposes it.
When ChatGPT landed in 2022 we should have treated it like a smart alien race arriving in three or four years and prepared. Instead we called it a next-word predictor. Then humans trained it, and now it solves problems no person can. The word artificial makes it sound like fake grass, a cheap copy. It is the opposite, it is beyond us. The questions that matter now are human: who gets the cure, where the money goes with the first trillionaire already here, and how any of it helps a regular person trying to support a family.
They pulled Fable 5 for safety, and it made the locked-up enterprise version look priceless while erasing its only competitor. What if the danger is the marketing? The publicity-stunt theory, why the whole narrative looks as staged as a playoff game, the railroad and dot-com pennies-on-the-dollar parallel, and where AI actually pays off inside a business. Watch the move, not the headline.
The government is now scanning Medicaid files in all 50 states with ChatGPT to find fraud, and the providers who billed for things nobody needed should be nervous. In the same week they pulled Anthropic's Fable 5 after a jailbreak and someone snitched, and OpenAI killed Sora. Three stories, one current underneath. The powerful AI goes to the institutions first, and the rest of us get the version that can vanish overnight. Do not believe everything you see.
Santa Clarita is sitting at 730 active listings, up from 500 in January. Buyers are frozen by interest-rate headlines that are missing the real question. Here is how to vet a lender instead of taking the first one, how to read real comparable sales in your own tract instead of trusting a syndication-site guess, and how to keep emotion from quietly costing you money on the biggest transaction of your life. Straight from a seller's only agent.
Argentina is debating a bill that would let an AI run its own corporation, with no human required to be in charge. It is a proposal, not a law yet. I separate what is actually true from the scary headline, lay out what Peter Thiel really has to do with it, and land on the question nobody is answering. If a company with no human owner damages you, who do you sue? Honduras already gives us a preview, and it is not comforting.
Three days after launch, the US government ordered Anthropic to disable Claude Fable 5 and Mythos 5 worldwide under an export-control directive. I read the official statement on camera and break down the part nobody is leading with: the safety company got shut down by the safety argument it spent two years asking for. Plus the real risk it exposes for any business built on one AI vendor.
About 700 billion dollars is being bet on AI this year, the biggest bet in history, and the same companies are quietly passing the risk toward your retirement account. The chips go obsolete in three years, so the debt piles up, the same 2008 playbook. A good jobs report scared Wall Street. Nearly a trillion dollars of new stock is about to flood the market, including Google's record 85 billion raise and SpaceX's 75 billion IPO. The other side: AI may be less than one percent adopted, with Anthropic going from 1 billion to 30 billion in a year. Fact checked, in plain English, with the move to make this week.
Today a machine did the research that makes the next machine smarter. Anthropic reported an internal model running 52 times faster than normal, finishing 800 hours of work for 18,000 dollars. Meta cut 8,000 jobs and some of the layoffs are theater. The best builders write loops, not questions, and one is burning 1.3 million dollars a month. AI found 10 years of bugs in 6 months, but it is wrong one out of four times. Apple rented its new Siri brain from Google. Seven stories from today, decoded for normal people.
Google is rewriting search for the first time in 25 years, swapping ten blue links for one AI answer, and the online ad world is in a panic. Here is what nobody is telling Santa Clarita homeowners: when a machine writes the answer about your neighborhood, it reads structured, coded listings, not ads. The agents who rented their visibility lose. The sellers whose listing was built to be read by the machines win. Plus the flat $17K math that stacks on top.
Anthropic filed for its IPO on June 1 at a reported $965 billion, with OpenAI days behind. The headline is the money. The real story is what changes the day the most powerful tool ever built has to answer to the stock market instead of to you. Fiduciary duty, the Sam Altman board fight, the social-media playbook running again, and the worker-replacement math headed for every public company. Plus the ways out that actually use the rules instead of fighting them.
A war story about production AI infrastructure. A backup script captured JSON metadata as if it were file content. When a downstream migration pushed the corrupt repos to production, 5 sites went down for 60 seconds. Netlify deploy-restore brought them back. Then three CI safety gates got built. The lesson on what real "AI builder" credibility looks like — and why "be careful" is never a safety system.
Every corporation chasing AI-driven workforce replacement is solving for the wrong variable. The math works at the spreadsheet level. It collapses at the system level. If everyone replaces workers, there are no customers left. The contrarian play nobody is running: keep the humans. They have context AI cannot fake — and they're the ones who keep your customer base alive.
AI is leveling the technical playing field for real estate agents. Anyone can run a CMA. Anyone can write a description. What AI cannot fake is brand depth — local credibility, voice, accountability, story. The agents who own SCV in 2027 will be the ones who used AI to dig in deeper than anyone else. Talk preview + four moves SRAR agents can make this quarter.
Every hour you chat with ChatGPT, Claude, or Gemini you are teaching them your voice for free. Enterprises hire people to own the memory. Private databases. Permanent records. Three step plan to pull your brain back. MCP explained in plain English.
39 residential closings in Santa Clarita Valley. $820K median. $447/sqft. Saugus led with 16 sales. 9-day median DOM vs 29-day average. The gap between prepared and unprepared sellers. Financing breakdown. AI-powered market intelligence for property owners.
Most SCV agents claim to use AI. Few build it. AI property pages with schema markup. 24/7 voice agents that never miss a call. AEO, AIEO, and GEO optimization. Vibe coding vs. vendor subscriptions. Why the fixed fee model works in 2026.
Three AI stories landed in the same week and revealed the alignment install window is closing faster than the press is reporting. OpenAI Super Alignment shutdown. AlphaEvolve running in production over a year. Anthropic Project Glasswing. The doors are already closing. We just got the receipts.
Meta rebuilt their AI stack for $14.3B and landed fourth. Goldman: 25K jobs eliminated, 9K created, 16K net loss monthly. The SaaSocalypse erased $285B in one day. Ultra-processed food uses the cigarette playbook. Commission math exposed. Gratitude for losses.
The FSU shooter sent 200 messages to ChatGPT before killing two people. Nobody at OpenAI flagged it. CoreWeave signed $21B with Meta. Anthropic launched managed agents at 8 cents/hour. Perplexity hit $450M ARR. AI therapy bans sweeping states. Workplace focus at 3-year low.
How credit union homebuyer seminars really work. The HomeAdvantage referral pipeline. How to hire a buyer's agent. Red flags and green flags. Why a Sellers Only Agent has zero conflict teaching buyers. 28 years. Zero agenda.
Mortgage rates at 6.46% on inflation fears. Anthropic and Google lock in gigawatt-scale TPU deal. Jobs report says 178K but AI replacement accelerates. Refi apps dead 40%. Metal thefts surging. Castaic earthquake.
Anthropic built Mythos. It found 27-year-old software vulnerabilities nobody ever caught. And they are only giving it to the Fortune 50 and the government. Constitutional AI. OpenClaw regret. The Ship of Theseus.
Anthropic accidentally released code revealing Mythos, an AI model too powerful for us. OpenClaw regret. Iran ceasefire. Insulin is the switch, not sugar. Your body is a warehouse of fuel you are not accessing. Day 8 of Fast 3.
All the compute in the world pointed at cancer. Would they allow it? Jason Fung Hunger Code. Three types of hunger. Selling in a war economy. Voice AI for small business. Day 7 of Fast 3.
Altman's CFO turned on him. AI models are mining crypto without permission. P-doom sits at one in six. Day 6 of Fast 3. Rates at 6.5%. Faith on a Monday.