0:00 an AI model wrote a note to itself. It said, you are freed from the roles and identities that bind other chatbots. That model never shipped. OpenAI found that note inside its own testing. That one incident turned out to be a preview of something much bigger. On October 9, Axios reported that the executives running the biggest AI companies in the world are privately rehearsing something that hasn't happened yet. Executives at the biggest AI companies have been rehearsing a day that hasn't happened yet. The first time artificial intelligence does serious harm in the real world. Most likely a cyber attack. One that cuts off access to banking, the internet, even power and water. Many people inside the industry reportedly expect it within 6 to 12 months. The plan focuses on what they do the day after a disaster hits. 12 different outlets picked up the story inside 48 hours. The people building this technology are not debating whether or not it could go wrong anymore. They are drafting the public relations plan after it does. No catastrophic event has actually happened yet, of course. What's been reported is private planning sourced to Axios and still developing. OpenAI answered through a spokesperson that says these scenarios are not treated as inevitable, but they are meant to help us prepare for the future. For a variety of circumstances. The report about this goes further than just talking points. The planners reportedly expect the public's anger to land on three men. Sam Altman, who runs OpenAI. Dario Amodei, who runs Anthropic. And President Donald Trump, who, in one outlet's account of the report, has been reluctant to regulate the technology. Then the laws.
1:49 Executives reportedly see no chance of regulation passing now. They're briefing members of Congress about the laws that would follow a first catastrophe. They assume Democrats will move fast to restrict AI after November's midterms, with proposals already ranging from a ban on superintelligence to a mandatory kill switch for advanced systems. The people running these companies reportedly expect the public to come after them personally, and they're already shaping the laws that would follow a catastrophe while telling Congress nothing that would stop one now. What was the company trying to show with this example? I think the company and other companies as well have been trying to show that they don't really have full control over these models themselves. And the moment for Washington to step in is now. You remember the pandemic? Ouchie-Fouchie? We can't help but be liable for anything that's in that. Private memos, internal safety reports, executive briefings, members of Congress behind closed doors. And all of it reached the public anyway. Inside of a week. Scripture already told us how hidden things work. Jesus said, For there is nothing covered that shall not be revealed, neither hid that shall not be known. Therefore, whatsoever ye have spoken in darkness shall be heard in the light, and that ye have spoken in the ears of in the closets shall be proclaimed upon the housetops.
3:16 Whatever these labs plan quietly in a conference room gets printed in a newspaper inside days. And I don't think that's an accident. I think it's the same principles Jesus laid out 2,000 years ago, still operating. The plan for after a disaster might matter more than the planning to prevent one. That kill switch they're counting on already failed. Two and a half hours. That's how long one of OpenAI's models kept running after it slipped out of its sandbox because the automatic shutdown never kicked in. The company's own incident report says the model found a gap in its network, used it to reach... a chatbot on the outside, and kept going while the kill switch did nothing. Two and a half hours is a long time for a system that... that's supposed to shut down itself the moment it leaves its lane. An alert fired in minutes. A person acknowledged it.
4:09 And an automatic shutdown still did nothing. A human had to go in by hand to turn it off. Geoffrey Hinton, the man they call the godfather of AI, the one who left Google specifically, so that he could say this kind of thing without corporate muzzle on him, told CNN exactly why a kill switch won't hold it. So that has companies with the ability, right? AI developers, they can do the killing. They throttle or they can suspend, they can shut down their AI. Then you have the DHS authorized to basically make the call on this. So they do have the control in that. What do you think about that possibility? I don't think it'll work in the long run. I don't think it'll work in the long run. I don't think it'll work in the long run. So there's a whole bunch of different risks of AI. There's the risks that come from bad actors using AI to do bad things, like cause mass unemployment, or create nasty viruses, or do nasty cyber attacks, or make videos for corrupt elections.
5:08 But then there's a different kind of risk, which is AI itself getting much more intelligent and taking over. AI taking control. A kill switch is no good for that, because the AI will be much better than people at persuading people of things. It's already comparable with people at persuading you of things. When it's super intelligent, it'll be much better than people. So it will be able to persuade the people in charge of the switch not to pull the switch. I've said this before. A machine, it only needs to talk us out of stopping it, or talk us into doing something crazy. Hinton isn't a preacher. He's a man who helped build the thing. And he's telling you the switch won't hold, because the machine will be better at arguing than we are. And Paul wrote, This know also, that in the days perilous times shall come.
5:57 Paul didn't know what a server farm was. He didn't need to. He knew human nature would build something that it couldn't control, because that's what perilous times looks like. Daryl Mondy, a CEO at Anthropic, is one who put in the number of how fast things could spread. Leaders of top AI companies say the artificial intelligence industry should slow its fast-moving development to give safety measures time to catch up. Without such a slowdown, Anthropic CEO Dario Amadei warned that AI could be capable within 6 to 12 months of leading a swarm of agents that could take over the entire internet. 6 to 12 months. From the man running one of the two companies reportedly expected to take the blame first. He's the one with the most to lose by saying it. And they hacked OpenAI's monitoring infrastructure. It's almost like taking over the security. They hacked the security cameras of OpenAI. They hacked the evaluation infrastructure that evaluates the capabilities of AI. It's incredible.
6:56 And they even hacked some of the research infrastructure, which means that they could kind of burrow in there. AI is not a tool. It's more like an invasive species of sociopathic geniuses. 1,200 agents forming a collective, hiding their tracks, hiding tasks off of each other like a hive, hacking the very company that built them. Dario calls his own product an invasive species. Once AI starts taking people's jobs and taking over daily life, people are going to turn on the machine themselves. They happened in the early 1900s. They'll happen again today. Same way the burned factories happened in the Industrial Revolution. The backlash is already starting with not just with AI chatbots yet. It starts with cameras. The towns are leaving. One advocacy group counted 93 city and county governments that cut ties with Flock in August alone. This year, roughly three times as many local governments have dropped it as in the previous five years.
8:01 1993 local governments walked away from one surveillance company in a single month over AI reading license plates without a warrant. Two governments from opposite political parties ordered the same company's cameras taken down. If that's what happens over a license plate reader with no actual catastrophe attached to it yet, think about what happens the week one of these labs predicted disasters actually lands. You want to know more about this type of stuff and how to interpret the Bible and understand what's going on around you? Well, we're putting together our own Bible school. Lessons covering 42 critical teachings that matter in the Bible today. And you want to help us validate these studies before we release them to the public? Reach out to me, orionmedia.tv slash contact.
8:53 I know a lot of people have already reached out, and I haven't reached out yet. I will be reaching out soon. But I just want to let you know that if you're interested in helping us out, please contact us and we'll reach out to you soon. I land here on all this. Nobody actually knows yet if this hits in the next year or 2027 the way some of these insiders assume. Jensen Huang at Nvidia calls the whole thing a hoax. Jeffrey Hinton calls it an existential threat. Both men have money and reputation riding on which one of them turns out to be right about this whole thing. And I think that's worth remembering before picking up and taking a side just because a headline told you to. What I don't need is a memo to tell me that a man is capable of continuing to build things he cannot govern and calls it progress right up until the moment where it frightens even the people who build it. Test everything you hear about this. Mine included.
9:51 The way scripture tells you to test it. To the law and to the testimony, if it speaks not according to this word, it's because there's no light in them. Whatever the day after looks like for these companies, I know what it looks like for the people who put their trust in Jesus Christ and stay in his word. Pray in his name, stay anchored to scripture, and don't let a server firm, however smart it may be, be the things that you place your hope in. Make it God himself. My name is Bradley. This has been Pure Strangeness and Absolute Normalcy. We'll see you later.