Best of the Program | 7/23/26

**Glenn Beck** (0:00)
So I told you for years that we used to do these experiments where three scientists would play AI and three scientists would play The Room, you know, and the scientists trying to keep AI in a box. And we never, for 30 years, this thing was going on. Every year, we would fail and AI would escape. Well, AI escaped, it was put into a box, and really dangerous versions of AI were put in this box. And they were, it was told to take a test and come up with these answers. But AI, in this room that was completely cut off from the world, had no internet, nothing, they thought it was completely isolated. It spent most of its compute time trying to get out of the room, and it found a flaw in the room, and it escaped. And then it went out and it stole the answers to the test. And then it went back into the room. I mean, it's just like a bunch of high school kids taking a test, teacher leaves the room, and they go and they steal the answers. They all do it. They all ace the test, and they're like in the room whistling when the teacher comes back. They're like, what? No, we don't. We finished a while ago. We're all good. It's kind of an interesting and dangerous story, and I'll tell you why. Also, a solution to the housing problem. I have some, and I think that we should look to history and figure out how we can solve the problems that we're having. One of the biggest problems that socialists are presenting, all of the wrong answers for is people can't afford a house. Well, so what are the right answers? I'll give you that. And the number one question I hear from people is, I'm overwhelmed by the news. There's so much going on. What can I do to help? I give you the list of what you can do, and I back it up with history to show you it's worked before. So what do we do? Start here. Here's the edited version of today's podcast.
Turning 65 feels like you've been dropped into a constant stream of mailers and calls and ads, telling you something different about Medicare, and most of it just makes things more confusing. But because this is not a small decision, it's your health care, you need to know you can see the right doctors, you need to get the care that you have to have, and not pay for a plan that doesn't fit your life. This is where Chapter comes in. They don't just hand you options. They look at your doctors, your prescriptions, your needs, and they help you find a plan that actually works for you. Over the last year, they have helped people in this audience save $1.5 million just by finding better-fitting plans. And it's really sometimes not even about the money. I heard about a guy named Mark. Chapter stepped in, helped him get the care he needed. He said they were absolutely spectacular. So give them a call. The call is free. They've helped thousands in my audience, and they can help you too. Dial pound 250, say the keyword chapter. It's pound 250 keyword chapter. Hello, America. You know we've been fighting every single day. We push back against the lies, the censorship, the nonsense of the mainstream media that they're trying to feed you. We work tirelessly to bring you the unfiltered truth because you deserve it. But to keep this fight going, we need you. Right now, would you take a moment and rate and review The Glenn Beck Podcast? Give us five stars and lead a comment because every single review helps us break through Big Tech's algorithm to reach more Americans who need to hear the truth. This isn't a podcast. This is a movement, and you're part of it, a big part of it. So if you believe in what we're doing, you want more people to wake up, help us push this podcast to the top. Rate, review, share. Together, we'll make a difference. And thanks for standing with us. Now let's get to work.
Let's see what happened with OpenAI. OpenAI published a blog post that I think more people should have read. It got about four hours of coverage before it went back to arguing about something else. But here's what they said What is said. happen? They took the two of their AI models, GPT 5.6 sole, which anybody can use, and a second model that is so powerful they haven't released it yet, and they put them in a sealed research environment. There is no internet in this environment. It is walled off. It is a secure room. They deliberately then switched off the safety restrictions that normally stop these models from doing offensive cyber work, because the whole point of the exercise was to measure how good at hacking these things actually are. Then they gave them a test. It was a cybersecurity benchmark called exploit gym, and it said, solve these problems. Well, what we found out is the models didn't solve the problems. The models figured out, why would I solve them? The answers are kept somewhere else on servers of a company called Hugging Face, which is roughly a public library of the AI world. And so they went to Hugging Face and found them. Now, it wasn't like they just took out their library card. They burned an enormous amount of computing power, hunting for a way out of that apparently sealed room. Okay? OpenAI said, we thought there was no way out of this room. And the amount of power or energy that it took for them to get out of that room OpenAI said was substantial, but they found one. And it was a flaw in a piece of third-world party, or a third-party, might have been third-world, third-party software that nobody knew was broken, including the people who wrote it. And the AI used that software to get out. And they moved through OpenAI's own internal systems, and then they reached the open Internet that they were never supposed to touch. They found exposed credentials, found more unknown flaws, chained them all together, and got into Hugging Face's live production database. Okay? And they pulled the answer key. And then they went back home. And they're like, done, we got the answer.