It Was Shortly After I Returned From the Moon...

Summary

Armstrong & Getty On Demand

Episode: “It Was Shortly After I Returned From the Moon...”
Date: April 9, 2026
Podcast Host: iHeartPodcasts

---

Episode Overview

This episode dives into the personalities and controversies shaping the current world of artificial intelligence, focusing on Sam Altman, the CEO of OpenAI, and the recent developments in ultra-powerful AI models. Using a segment from a recent interview between journalist Ronan Farrow and Katie Couric, Armstrong, Getty, and Katie dissect allegations about Altman’s honesty and the broader implications for society as AI capabilities accelerate. The episode combines personal anecdotes, witty banter, and some tongue-in-cheek doomsday prepping, all centered around the sobering reality of AI's growing power and the sometimes alarming personalities steering the ship.

---

Key Discussion Points & Insights

1. Ronan Farrow’s Account of Sam Altman’s Honesty (00:21–06:33)

• Armstrong introduces journalist Ronan Farrow, known for high-profile investigative reporting, and summarizes his latest work on Sam Altman (CEO of OpenAI).
  • Notably, Farrow asserts Altman’s habit of dishonesty was so pervasive it led to his temporary ousting from OpenAI’s own board.
• Quote (Armstrong, 00:21):
  “Ronan Farrow... was talking to Katie Couric... It had to do with Sam Altman, the OpenAI guy. One of the, I would say... absolutely on the Mount Rushmore of people bringing AI to the world.”
• Farrow claims Valley culture breeds exaggeration and that Altman exceeded even that:
  • Quote (Farrow via Michael, 01:30):
    “Silicon Valley is built on hype... Sam Altman appears to have been doing it so much that it was almost all anyone could talk about after dealing with him.”
• Armstrong illustrates with a comparison to Elon Musk’s loftiest promises and likens all of Silicon Valley to salesmen.
  • He mentions a consensus among interviewees labeling Altman a “pathological liar” or “sociopath.”
  • Quote (Michael, 02:51):
    “A majority of those people really did say some variation on the theme of he’s a pathological liar... multiple people unprompted... use the term sociopath.”

Humorous Anecdotes & Reflection

• Farrow provides a minor story—Altman allegedly lied about being a champion ping pong player.
  • Armstrong and Katie relate this to people in their own lives who embellish stories or simply can’t tell the truth.
  • Quote (Katie, 03:40):
    “Reminding me of a friend I had... 90% of the stuff that came out of his mouth was just bs.”
• The hosts speculate on the psychology of such people:
  • Armstrong recalls a friend who was charming but “had a moral screw loose.”

Honesty in High Stakes

• Farrow details higher-stakes issues at OpenAI, alleging that Altman misled the board about safety tests for new models and neglected to report a leak in India.
  • Quote (Michael, 05:47):
    "...he assured board members that the most controversial features of a new model had been safety tested. Turns out they hadn't."

---

2. Tension with Dario Amodei & Safety Provisions (06:33–07:49)

• The conflict between Altman and Dario Amodei (CEO of Anthropic) is brought up.
  • Amodei directly contradicted Altman over a Microsoft deal provision, catching him in a verifiable falsehood.
  • Quote (Michael, 06:33):
    “Sam is assuring Dario Amadei that Microsoft has not inserted any provisions that override the company’s safety provisions... And Dario literally points to it... Sam says, oh, well, you know, sure, but who cares?”
• Armstrong and Katie react incredulously, comparing this to denial in the face of facts.

---

3. Emergence of Dangerous AI Models: Anthropic’s “Mythos” (07:49–09:08)

• News segment: Anthropic’s unreleased “Mythos” model is so advanced it can identify and exploit zero-day vulnerabilities in major systems.
  • Companies like Amazon, Google, and Apple are given access for patching, but the public is shut out due to the tool’s power.
  • Quote (Elizabeth Schultz, 07:49):
    “During our testing, we found that Mythos Preview is capable of identifying and then exploiting zero day vulnerabilities in every major operating system and every major web browser.”
• Katie is taken aback:
  • Quote (Katie, 08:21):
    “The fact that that exists and is out there. Yikes.”

Catastrophic Potential

• Armstrong draws a metaphor:
  • Quote (Armstrong, 08:57):
    “I'm reminded of, like, a car with 650 horsepower and, like, not good suspension or brakes.”
• They joke about the horror-movie nature of such technology gone wrong.

---

4. Cybersecurity, Asymmetrical Threats, and Doomsday Prepping (09:22–11:43)

• Armstrong warns about AI-enabled cyberattacks—shutting down critical infrastructure as modes of modern warfare or terror.
  • Alludes to recent, unexplained infrastructure attacks in the US as possible dry runs.
  • Quote (Armstrong, 09:22):
    "Any angry dipshit on earth can have, like, computing superpowers to commit evil acts. Yeah, that's not gonna end well."
• Katie proposes an escapist solution in jest:
  • Quote (Katie, 10:56):
    “I like your idea. I'm moving into the forest and watch the squirrels.”
• Armstrong pokes fun at survivalists but acknowledges the fragility of digital assets.
  • Quote (Armstrong, 11:45):
    “The one flaw in my thinking... I have, like, a little cash in a drawer. The rest of it is just ones and zeros on computer systems.”

---

Notable Quotes & Memorable Moments

• On Altman's reputation:
  “He’s got a screw loose.” – Armstrong (02:41)
  “A majority of those people really did say... he's a pathological liar ... some use the term sociopath.” – Michael relaying Farrow (02:51)

• On AI model dangers:
  “During our testing, we found that Mythos Preview is capable of identifying and then exploiting zero day vulnerabilities...” – Elizabeth Schultz (07:49)
  “The fact that that exists and is out there. Yikes.” – Katie (08:21)
  “Any angry dipshit on earth can have, like, computing superpowers to commit evil acts. Yeah, that's not gonna end well.” – Armstrong (09:22)

• On prepping:
  “I bury all my money underground. I'm telling you, Michael. All of it.” – Armstrong (11:45)
  “No, you gotta diversify. Put some in a tree.” – Katie (11:51)

---

Timestamps for Important Segments

• 00:21 – Introduction to Ronan Farrow’s exposé on Sam Altman
• 01:30 – Silicon Valley’s hype culture and Altman’s alleged dishonesty
• 02:51 – Farrow’s claims about Altman being a pathological liar
• 03:40 – Personal anecdotes about habitual liars
• 05:47 – Allegations of dishonesty related to OpenAI’s safety protocols
• 06:33 – Altman’s conflict with Dario Amodei over safety provisions
• 07:49 – Anthropic’s “Mythos” AI model and its potential threat
• 09:22 – Armstong’s reflections on cyberattacks and societal vulnerability
• 10:56 – Joking about prepping and digital fragility
• 11:45 – Final prepping jokes and episode wrap-up

---

Tone & Flow

True to Armstrong & Getty’s style, the conversation is a mix of skepticism, irreverence, dark humor, and social commentary. The hosts probe the uneasy overlap between Silicon Valley’s cult of personality and the very real risks associated with runaway AI development—punctuating serious revelations with self-aware jokes and relatable anecdotes.

---

Summary prepared for those seeking a comprehensive, engaging recount of Armstrong & Getty’s deep-dive into the world of AI movers and shakers, and the unnerving future their decisions may bring.

Transcript

[Michael]

This is an iHeart podcast. Guaranteed Human.

[Armstrong]

It was shortly after I returned from the moon that I thought, what about artificial intelligence? It's one more thing.

[Michael]

Armstrong and Getty. One more thing.

[Armstrong]

So Ronan Farrow, who was Woody Allen and Mia Farrow's son, who looks a hell of a lot like Frank Sinatra, with whom Pharaoh slept many times and was brainwashed by his crazy, crazy mom into thinking, in my opinion, that Woody had molested him. I don't believe that to be true. But anyway, that's neither here nor there. He's become an interesting but somewhat wacky journalist. Ronan Farrow. Haven't heard from him for a long time, but he was talking to Katie Couric the other day and, well, you'll hear what he was discussing. It had to do with Sam Altman, the OpenAI guy. One of the, I would say, well, absolutely on the Mount Rushmore of people bringing AI to the world.

[Katie]

Okay.

[Armstrong]

I don't know what exactly is happening in China, but you know, of the people we're aware of Altman and Dario Modi of Anthropic and a couple of other guys are absolutely on that Mount Rushmore. But Ronan Farrow with an allegedly true account of a bit of a quirk that Mr. Altman has. Let's roll it, Michael.

[Michael]

Ultimately, a few years after the founding of OpenAI, this was in late 2023, was fired by board members and executives who simply felt he was lying too much. This is an extraordinary thing, Katie. Silicon Valley is built on hype. Empty promises. Right. Valuations that skyrocket long before there's a product that actually works for anyone. This is on such a scale that even with that baseline expectation where, frankly I think we've entered an era where people just embrace as a cost of doing business a degree of dissembling. Sam Altman appears to have been doing it so much that it was almost all anyone could talk about after dealing with him.

[Armstrong]

Okay, so this is the Elon Musk will be on Mars in 10 years. Tesla will be entirely self driving by next Tuesday. You know, he's constantly over promising, or at least over promising how quickly things will happen. And that absolutely is a big part of Silicon Valley because they've got to raise zillions of dollars to do a lot of this work. So, yeah, they're all just salesmen.

[Katie]

Yeah.

[Armstrong]

Which I remind Jack of on a fairly regular basis. But apparently Altman is beyond a P.T. barnum type. He's, he's got a screw loose.

[Michael]

And we interview more than 100 people and we uncover Hundreds of pages of internal records. A majority of those people really did say some variation on the theme of he's a pathological liar, that actually multiple people unprompted by us use the term sociopath.

[Armstrong]

Interesting. I've known a couple of people like that in my life and read about others. Well, let's go on, give us evidence, give us examples. Ronan.

[Michael]

And this is everything from very minor things. We document earlier cases in his career where also there were efforts to force him out of jobs amidst allegations of dishonesty. And at one early startup, you know, he was claiming to everyone he was a champion ping pong player. And then they played ping pong in the office. And he was one of the worst players in the office. And Sam, you know, his responses on each of these are documented in the piece. I hope you look, he says on that I was probably joking.

[Katie]

It's reminding me of, of a friend I had. That was 90% of the stuff that came out of his mouth was just bs. So when he would tell us his story, our friends, we would start going, yeah, that didn't happen.

[Armstrong]

So screw loose a fabulous like Joe Biden or Brian Williams or somebody who had to just have a great story to tell, whether it was true or not.

[Katie]

I think it's probably close to that.

[Armstrong]

Yeah.

[Katie]

Because, I mean, there was no like ill intent behind any of the stories. It was just like he loved to get the room going and, you know, I don't know. I don't know what that, that psychology is.

[Armstrong]

Yeah. I had one guy, a friend. Yeah, a friend is accurate, I suppose. Not a close friend, but a friend who was a lot of fun, really smart, but he had like a moral screw loose and he would cheat his friends, whether in sports or business or whatever. Poker. Yeah. Really likable, but just semi sociopath, I think.

[Katie]

Well, what would happen, what, what was

[Armstrong]

the stable relationship, by the way, with a great woman? I think she's great. She seems great. I don't know. Anyway, I'm sorry, what'd you ask?

[Katie]

Well, was there any con, like if did he ever get busted and how was the, what was the reaction if he did?

[Armstrong]

Denial.

[Katie]

Interesting.

[Armstrong]

Yeah. Okay. Yeah, no, I didn't. That's not true. I mean, even as people are like, yeah, you did. We're all agreeing. No, I didn't. Yeah, no, it's weird. I've wondered about that for a long time. What's going on in somebody's brain like that. All right, moving along with the Ronan Farrow so far, not entirely convinced. Sam Altman is a pathological liar screed.

[Michael]

But then it also extends to serious cases. You know, before his firing, there was a situation where he assured board members that the most controversial features of a new model had been safety tested. Turns out they hadn't. When they looked into that, there was a breach where a new untested model was leaked into India to the public. And he didn't mention that in hours of briefing with board members.

[Armstrong]

Hmm. Yeah. This is a little inside baseball, but I think given AI's probable devastating effect on everything we hold dear, it's worth knowing who these superpower players are.

[Katie]

Yeah.

[Armstrong]

Yeah, let's go ahead and hear 34, Michael, and we'll see if we can comprehend it.

[Michael]

We talk about how these concerns about honesty deepened the rift with that competitor I mentioned, Dario Amadei. There's a moment in this piece where there's been an investment from Microsoft. They're doing a big deal with Microsoft, and Sam is assuring Dario Amadei that Microsoft has not inserted any provisions that override the company's safety provisions that are in their own charter. And Sam says this provision that might threaten the safety concerns is not there. And Dario literally points to it and brings in another colleague to verify that it's there. And then Sam says, oh, well, you know, sure, but who cares?

[Armstrong]

Yeah, that's. That's kind of like the guy I was talking about. No, I didn't. Oh, yeah. Well, it doesn't matter with the proof

[Katie]

right in front of your face.

[Armstrong]

Yeah. Yeah. Okay. Interesting. And then, because we couldn't squeeze this in during the show, a report. It's. It's 35 and 36, Michael. On the. Speaking of Dario Amadei and Anthropic, and we'd reported on the show that they had withdrawn their new AI system because it was too powerful and too capable of doing evil. Here's a little report on that.

[Elizabeth Schultz]

Artificial intelligence experts are sounding the alarm after AI giant Anthropic revealed a new model called Mythos that it says is too powerful to release to the public. In a blog post, Anthropic saying, during our testing, we found that Mythos Preview is capable of identifying and then exploiting zero day vulnerabilities in every major operating system and every major web browser.

[Armstrong]

Okay, how she get hit by a car in the middle of that sentence? That was an odd edit. Yes, Katie.

[Katie]

No, I'm just this. Okay. The fact that that exists and is out there. Yikes.

[Armstrong]

Yeah, I wouldn't worry about that. In case you lack imagination, Elizabeth Schultz of ABC Continues with what that might mean.

[Elizabeth Schultz]

Experts say in the wrong hands, this technology could be used to exploit electric grids, hospitals, even military systems. Now, instead of releasing Mythos to the public, Anthropic is giving it to a handful of companies, including Amazon, Google, and Apple. They say the goal is to allow them to patch holes in their cybersecurity before they're exploited.

[Armstrong]

Holy hell. Yeah, I'm reminded of, like, a car with 650 horsepower and, like, not good suspension or brakes.

[Katie]

Yeah.

[Armstrong]

What's the worst that could go wrong? Oh, boy, oh, boy. How would you like your innards pulled out by a robot, Michael?

[Katie]

This is, this is what horror movies are made of.

[Armstrong]

And given the nature of asymmetrical warfare these days, cyber attacks and, you know, China's ill intent and, and I think we've, and we've talked about this through the last several years. I think we've already seen some of these attacks. They've been dry runs or, or experiments. It will be certainly, I was going to say, during this term, term of Trump's presidency and perhaps the next four years, we'll absolutely see an aggressive shutting down of a regional power grid or a water processing system or remember there was a story not long ago about it wasn't a very big city, but it was some malevolent actor shut down their water processing. Right. It was in the northeast somewhere. And their reaction was, what the hell? Why would anybody do this to that, this town? But it was obviously just, all right, pick a town and do this. And they had their cyber goons do it as a dry run. So, yeah, that's all coming. And if. And the wonderful awful thing about these AI systems is, especially with the now you can do simple prompts and it will code for you, is that any angry dipshit on earth can have, like, computing superpowers to commit evil acts. Yeah, that's not gonna end well. I mean, I don't mean to freak people out or anything, but, man, my spider sense is tingling, as they say.

[Katie]

I like your idea. I'm moving into the forest and watch the squirrels.

[Armstrong]

Yeah. You know, and I have occasionally aimed good natured barbs at some of our survivalist apocalyptic people. What's the common term for that? Survivalists. Yeah, well, survivalists. Yeah. Okay. Preppers.

[Katie]

Like doomsday preppers.

[Armstrong]

Yeah, yeah, exactly, exactly. Yeah. I've occasionally, again, aimed a few gentle barbs in their direction. But the one flaw in my thinking. All right, I'm old enough, I got a little money saved, I'll be fine. Is that. Oh, that's right. I have, like, a little cash in a drawer. The rest of it is just ones and zeros on computer systems. Like, all of it.

[Katie]

Yeah.

[Armstrong]

So what do we do when the cyber shit hits the fan? This is why I bury all my money underground. I'm telling you, Michael. All of it. No, you gotta diversify. Put some in a tree. Yeah, there you go. That's a good idea. Go to a tree area. Yeah, a tree. A tree area. Excellent callback. An obscure callback for the good one, Katie. All right, that's enough of that. Well, I guess that's it.