Latest / Connor with Honor: AI, Real Estate and Real Life / AI Daily Show 9/3/2026: The Fear Dividend, Who Gets Paid When You're Scared of AI

Transcript

AI Daily Show 9/3/2026: The Fear Dividend, Who Gets Paid When You're Scared of AI · Connor with Honor: AI, Real Estate and Real Life

• 0:00
  SPEAKER_00: On Wednesday, 38,000 people got hired in the private sector of the United States, and that's the whole month of August, and that's the weakest month since January.
• 0:09
  SPEAKER_00: And that same week, three companies announced models that can break into hardened computer systems without a human touching the keyboard.
• 0:17
  SPEAKER_00: And then told us we can't have them.
• 0:20
  SPEAKER_00: Those two things sound unrelated, but they're actually the same story, and both sides of an AI argument that are going to try to sell you fear about it by Friday.
• 0:29
  SPEAKER_00: Good morning, good afternoon, or good evening, whenever you happen to be tuning in.
• 0:32
  SPEAKER_00: Today's AI update is brought to you by Santa Clarita artificialintelligence.com, the practical AI portal for Santa Clarita business owners.
• 0:40
  SPEAKER_00: Everything here is dated and sourced, and where a company is talking about itself, I'm going to say so out loud.
• 0:46
  SPEAKER_00: Three things this morning, OpenAI, Anthropic, and Google all cross the same line in the same week.
• 0:52
  SPEAKER_00: And what they did about it tells you more than the models do.
• 0:56
  SPEAKER_00: The Pentagon just handed three million people to chatbots built for, in their words, warfighter needs.
• 1:04
  SPEAKER_00: And the job numbers came in soft while a bill landed in Congress to tax AI if it eats the labor market.
• 1:10
  SPEAKER_00: The practical move at the end is not to buy an agent, it's to find the switch.
• 1:15
  SPEAKER_00: Hold on to number one for me, one number, 1949.
• 1:18
  SPEAKER_00: Not a model number, but a year.
• 1:20
  SPEAKER_00: I'll come back to it in about seven minutes, and it's going to reframe every AI job headline you read this fall.
• 1:26
  SPEAKER_00: Start with the labs because they move together, and this is rare.
• 1:29
  SPEAKER_00: OpenAI published a piece called Path to Astra.
• 1:32
  SPEAKER_00: Astra is their next model, and it's the first one they say crosses the critical cybersecurity threshold in their own preparedness framework.
• 1:40
  SPEAKER_00: Plain English and expert testing the model found previously unknown security holes in a hardened browser and hardened operating system.
• 1:50
  SPEAKER_00: Chained them together and escaped the chat box and rand commands on the host machine.
• 1:56
  SPEAKER_00: Nobody guided it step by step.
• 1:58
  SPEAKER_00: That's the company's own claim about its own model.
• 2:01
  SPEAKER_00: Company claim.
• 2:03
  SPEAKER_00: Treat it that way.
• 2:04
  SPEAKER_00: But here's what makes it credible.
• 2:06
  SPEAKER_00: They're not selling it to you.
• 2:08
  SPEAKER_00: The most advanced cyber capabilities are going to a small group of testers first, then to defenders through a program they call Daybreak.
• 2:16
  SPEAKER_00: Same week, Anthropic Ship Clawed, Fable 5.1, and Claude Mythos 5.1.
• 2:21
  SPEAKER_00: Fable is the one anybody can use.
• 2:23
  SPEAKER_00: Mythos is the one you can't unless you're inside a trusted access program built for cybersecurity and life sciences.
• 2:30
  SPEAKER_00: Their published line is that Fable 5.1 can find software vulnerabilities, not easy to say, but is not allowed to write the exploit.
• 2:39
  SPEAKER_00: Google did the same dance with a model called Gemini 3.8 Flash Cyber, offered to trusted defenders through something called the Fairwind Program.
• 2:48
  SPEAKER_00: So read that as a pattern, not three press releases, but three competitors in one week, each said the same sentence.
• 2:55
  SPEAKER_00: Our model can go do the testing that gets people arrested.
• 2:58
  SPEAKER_00: And we're going to lock the door and hand the keys to people we've at.
• 3:01
  SPEAKER_00: You know, I spent a career around locked doors when three people who hate each other agree on where the fence goes, the fence is real.
• 3:09
  SPEAKER_00: And that's not doom, that's the opposite.
• 3:11
  SPEAKER_00: It's the first time the labs have behaved like a licensed contractor instead of a guy with a truck.
• 3:15
  SPEAKER_00: Now the part for the shop owner.
• 3:17
  SPEAKER_00: None of this touches your business directly.
• 3:19
  SPEAKER_00: What touches your business is the phishing email that gets written by the same class of tool.
• 3:24
  SPEAKER_00: And the vendor who tells you their AI is secure because the brochure sheds so.
• 3:29
  SPEAKER_00: The question is to ask any vendor this fall is simple.
• 3:33
  SPEAKER_00: Who can this thing act as and what can it touch?
• 3:36
  SPEAKER_00: And if they can't answer in one sentence, you have your answer.
• 3:40
  SPEAKER_00: Second story, the Department of War, which is what the Pentagon calls itself this year, turned on ChatGPT, MIL, and GROC for government inside of a platform called genai.mil.
• 3:52
  SPEAKER_00: More than 3 million military and civilian personnel now have access alongside Google's Gemini, which was already there.
• 3:59
  SPEAKER_00: 1.7 million unique users have logged in.
• 4:02
  SPEAKER_00: The stated use is unclassified work, planning, policy, logistics, framework, how long to microwave that burrito.
• 4:09
  SPEAKER_00: Confirmed by the department and every outlet that covered it.
• 4:12
  SPEAKER_00: Here's the analogy that fits.
• 4:14
  SPEAKER_00: Now it's not Skynet.
• 4:15
  SPEAKER_00: This is the world's largest employer running out the world's largest word processor upgrade, and it's doing it with three vendors, so no single one owns the building.
• 4:25
  SPEAKER_00: If you run a business with six people, the lesson is the same one the Pentagon just paid for.
• 4:30
  SPEAKER_00: Do not marry one model.
• 4:31
  SPEAKER_00: Put the tool behind a door you control, decide what it's allowed to see, and keep the humans on the decisions and carry a signature.
• 4:40
  SPEAKER_00: The one detail worth remembering a federal judge ruled last week that the same Pentagon's earlier move to blacklist anthropic was illegal and baseless.
• 4:49
  SPEAKER_00: So the government that banned one lab in the spring is buying from two others by fall.
• 4:55
  SPEAKER_00: This is not a conspiracy that is procurement and it's messy and it's political, and it means nobody at the top has figured out either.
• 5:03
  SPEAKER_00: Now the jobs, because this is a story that runs the fear economy on both sides, and I want to take it apart carefully.
• 5:10
  SPEAKER_00: The facts first.
• 5:11
  SPEAKER_00: Private payrolls rose$38,000 in August according to the ADP.
• 5:17
  SPEAKER_00: Below the$48,000 economists expected.
• 5:20
  SPEAKER_00: The weakest reading since January.
• 5:22
  SPEAKER_00: Job openings came in at$7.3 million flat.
• 5:25
  SPEAKER_00: Today at 8 30 Eastern, the government prints the August report.
• 5:29
  SPEAKER_00: And the consensus is somewhere between$25,000 and 58,000 jobs with unemployment holding at 4.1%.
• 5:37
  SPEAKER_00: Those are the numbers.
• 5:38
  SPEAKER_00: Nothing in them says AI.
• 5:39
  SPEAKER_00: Nothing in them rules it out.
• 5:41
  SPEAKER_00: Then the AI numbers trackers, the AI number trackers count somewhere north of 165,000 workers affected by layoffs in 2026, where the company named AI.
• 5:52
  SPEAKER_00: Over March and April alone, that outplacement firm Challenger, Gray, and Christmas tie 36,831 announced cuts directly to artificial intelligence.
• 6:04
  SPEAKER_00: Oracle has cut 21,000 over the past year and says AI will likely mean more.
• 6:10
  SPEAKER_00: Block, Snap, Monday.com all name the technology in the memo.
• 6:14
  SPEAKER_00: The technology, these are companies and also claims too, every one of them.
• 6:20
  SPEAKER_00: And here's what I want where I want to be fair because both camps want you scared.
• 6:24
  SPEAKER_00: When a company says we're cutting 2,000 people because of AI, there are three possible truths.
• 6:30
  SPEAKER_00: One, the machine actually does the work now.
• 6:32
  SPEAKER_00: Two, the company needed to cut anyway, and AI is the most respectable word in the English language right now, for we spent way too much.
• 6:41
  SPEAKER_00: Three, they're cutting people to pay for the AI, the chips and the data centers and the job losses come before the productivity, but not after.
• 6:50
  SPEAKER_00: Most of the memos are number two and number three, wearing number one's jacket.
• 6:55
  SPEAKER_00: That does not make the person who got walked out any less unemployed, though.
• 7:00
  SPEAKER_00: It means the reason on the press release is not the actual reason.
• 7:04
  SPEAKER_00: The fear economy runs on that gap.
• 7:06
  SPEAKER_00: On one side, the lab CEOs who tell you that a machine is about to be smarter than everyone, and by the way, buy your enterprise plan before your competitor does, on the other side of the pulpit, the podcast, and the politician who tell you it's coming for your kids.
• 7:21
  SPEAKER_00: And by the way, subscribe, donate, and vote.
• 7:24
  SPEAKER_00: Bill Gates said recently that executives are privately very worried and publicly downplaying it because there's money on the line.
• 7:32
  SPEAKER_00: Well, that thread pulled 258 votes on Reddit, and he's probably right.
• 7:36
  SPEAKER_00: But notice what both sides need from you.
• 7:39
  SPEAKER_00: They need you to be afraid because a frightened person needs a savior, and a savior gets paid.
• 7:44
  SPEAKER_00: And it's not a new trick, it's the oldest one.
• 7:47
  SPEAKER_00: 1949, here it is.
• 7:49
  SPEAKER_00: In 1949, mathematician at MIT named Norbert Wiener, the man who coined the word cybernetics, wrote a letter to Walter Ruther, the head of United States United Auto Workers.
• 8:01
  SPEAKER_00: Now he told Ruther that automatic machines were coming for factory labor and that the union should get ahead of it before the machines arrived.
• 8:08
  SPEAKER_00: The next year he published a book called The Human Use of Human Beings and said the same thing to the public.
• 8:15
  SPEAKER_00: Nobody built the thing.
• 8:16
  SPEAKER_00: He was warning about for another 20 years.
• 8:18
  SPEAKER_00: And a year later, 1950, Alan Turing published the paper where he asked whether a machine could think and propose the test we still argue about.
• 8:27
  SPEAKER_00: So the two arguments you're hearing right now, it will take jobs and it might outthink of it.
• 8:33
  SPEAKER_00: All of us were born on paper 76 years ago before the first transistor computer shipped.
• 8:39
  SPEAKER_00: The fear isn't new.
• 8:41
  SPEAKER_00: What's new is that the machine finally showed up, and it turns out the people who were right about that direction were wrong about the actual speed, twice in both directions.
• 8:50
  SPEAKER_00: Which gets me to the bill.
• 8:52
  SPEAKER_00: A bill landed in Congress this week to tax AI tokens, the units these models build in and use the money to fund jobs if technology causes mass unemployment.
• 9:02
  SPEAKER_00: Fortune covered it, 34 comments on the thread, and most of them missed the point.
• 9:06
  SPEAKER_00: Whether that bill passes isn't the story.
• 9:09
  SPEAKER_00: It's the story that the government is now writing insurance policies against the thing labs say is coming.
• 9:16
  SPEAKER_00: When the insurance shows up, the risk is real to somebody, even if the timing is a guess.
• 9:21
  SPEAKER_00: Here's the release valve because the numbers don't say what the headlines say.
• 9:25
  SPEAKER_00: Unemployment is 4.1%.
• 9:27
  SPEAKER_00: Jobless claims are near historic lows.
• 9:29
  SPEAKER_00: 7.3 million jobs are open right now in a country of 170 million workers.
• 9:35
  SPEAKER_00: 165,000 AI named layoffs is about one in 1,000, and that's the loudest year on record.
• 9:42
  SPEAKER_00: Direction, up size, small, and the sentence I'll keep saying until it's boring, productivity doesn't have to become unemployment.
• 9:50
  SPEAKER_00: A company can take three times the output and use it to answer the phone faster, ship the thing it could never afford and put the humans on the work that needs a human.
• 10:00
  SPEAKER_00: Whether it does, that is a decision a person makes, not the machine.
• 10:04
  SPEAKER_00: Now the practical part, and this is the whole reason this show exists.
• 10:07
  SPEAKER_00: I sit down with a business, I walk the floor, and the first thing I do is not open a laptop, I ask questions.
• 10:13
  SPEAKER_00: How does a lead come in?
• 10:15
  SPEAKER_00: What happens in the first 10 minutes after it does?
• 10:18
  SPEAKER_00: Who touches it?
• 10:19
  SPEAKER_00: Where does it sit over the weekend?
• 10:20
  SPEAKER_00: Where does it leak?
• 10:21
  SPEAKER_00: Nine times out of ten, the leak is not a place that needs an intelligent machine making decisions.
• 10:27
  SPEAKER_00: It needs a switch, a lever.
• 10:29
  SPEAKER_00: And this is the end that.
• 10:32
  SPEAKER_00: Then that, you know, if this, then that, this sort of thing.
• 10:35
  SPEAKER_00: The phone rings after five and nobody answers.
• 10:38
  SPEAKER_00: And that's a switch.
• 10:39
  SPEAKER_00: The estimate goes out and nobody follows up on day three.
• 10:42
  SPEAKER_00: That's a lever.
• 10:43
  SPEAKER_00: You don't hand a decision to an agent when a rule would do, because a rule does exactly what you told it.
• 10:49
  SPEAKER_00: An agent does what it thinks you meant.
• 10:52
  SPEAKER_00: Where the agent belongs is the step that actually needs judgment.
• 10:55
  SPEAKER_00: And only after you've written down what your judgment is, your philosophy, your positioning.
• 11:01
  SPEAKER_00: And you will not, you know, what you will and will not do for money because an AI agent in production is going to make a thousand small decisions in your name this month, and every one of them is going to sound like you, whether it's you or not.
• 11:13
  SPEAKER_00: That's the difference between the small shop and the giant right now.
• 11:18
  SPEAKER_00: The giant has 40 people who agree before the switch gets flipped.
• 11:23
  SPEAKER_00: The owner operator can see the leak on Tuesday and have the switch installed by Thursday.
• 11:27
  SPEAKER_00: Small is not the disadvantage this year.
• 11:29
  SPEAKER_00: Slow is.
• 11:30
  SPEAKER_00: And no, this is not the moment you call an agent into existence, tell it to run the business and go fishing.
• 11:36
  SPEAKER_00: I get asked that every week.
• 11:38
  SPEAKER_00: Maybe not about fishing, but you know, doing something fun.
• 11:41
  SPEAKER_00: It's not there.
• 11:42
  SPEAKER_00: Will it be?
• 11:42
  SPEAKER_00: I don't know.
• 11:43
  SPEAKER_00: And anybody that tells you they do is selling something in either direction.
• 11:47
  SPEAKER_00: What is there is a machine that will take the repetitive 60% off your desk if you tell it exactly where the fence is.
• 11:55
  SPEAKER_00: And that's the whole audit.
• 11:57
  SPEAKER_00: I find the leaks, install the switches, fence the gate, keep the signature.
• 12:01
  SPEAKER_00: One more thing from the boards this week because it said something about where the public actually is.
• 12:07
  SPEAKER_00: A survey making the round says one in 10 Americans thinks AI is conscious.
• 12:11
  SPEAKER_00: 223 comments arguing about it.
• 12:14
  SPEAKER_00: Meanwhile, the most upvoted post in the same community was a meme.
• 12:18
  SPEAKER_00: And the second was South Korea giving its entire population free access to AI with no token limits.
• 12:24
  SPEAKER_00: Conscious machines, free machines, and a joke all on one page.
• 12:29
  SPEAKER_00: That's where the country is.
• 12:30
  SPEAKER_00: Impressed, unimpressed, scared, and bored at the same time.
• 12:34
  SPEAKER_00: Your customers are in that same page.
• 12:37
  SPEAKER_00: Meet them there, not in the keynote.
• 12:39
  SPEAKER_00: So here's the move for today.
• 12:41
  SPEAKER_00: Pick one thing in your business that happens after you go home.
• 12:45
  SPEAKER_00: A form, a phone call, a message.
• 12:47
  SPEAKER_00: Write down what should happen in that first 10 minutes, and then ask whether that's a rule or a decision.
• 12:55
  SPEAKER_00: If it's a rule, build the rule this week.
• 12:58
  SPEAKER_00: It costs almost nothing and it never gets tired.
• 13:01
  SPEAKER_00: If it's a decision, write down how you would decide it and then let a machine try with you reading over its shoulder for the first 30 days.
• 13:09
  SPEAKER_00: That's not a strategy deck.
• 13:11
  SPEAKER_00: That's a Tuesday.
• 13:12
  SPEAKER_00: The machine didn't show up smarter than the warnings.
• 13:16
  SPEAKER_00: It should show up slower than the warnings and faster than all the plans.
• 13:19
  SPEAKER_00: Every fair on the market this week, from the lab, from the pulpit, is a bid for your attention with an invoice attached.
• 13:26
  SPEAKER_00: Keep the attention, read the invoice.
• 13:28
  SPEAKER_00: Artificial intelligence is coming fast.
• 13:30
  SPEAKER_00: We don't freeze on the shoulder.
• 13:32
  SPEAKER_00: We don't sprint into traffic because somebody yelled disruption.
• 13:36
  SPEAKER_00: We measure the speed, we understand the distance, and we make the move that gets everybody home.
• 13:41
  SPEAKER_00: AI for everyone, not just the wealthy.
• 13:43
  SPEAKER_00: I'm Connor with Honor.
• 13:45
  SPEAKER_00: Be safe out there.
• 13:46
  SPEAKER_00: We'll see you tomorrow.