How the AI Apocalypse Could Unfold as Tech Billionaires Prepare for Survival

Former US Air Force intelligence officer Drew Miller gestures toward a poster behind his desk showing different types of ammunition, pointing to what he believes could become a vital defense if supercharged artificial intelligence ever turned against humanity.

‘If AI comes after us, we will fight back,’ he told the Daily Mail. He singled out a 10-gauge shotgun, saying its range of more than 20 yards could help people target drones sent against them. ‘We want to be able to shoot drones as far out as possible,’ he said.

‘Just one of the little pellets, if it hits the blade, odds are it’s going to take it down,’ Miller added.

Miller’s warning centers on the nightmare possibility of Artificial General Intelligence, or AGI: a still-theoretical form of AI that would exceed human intelligence and potentially decide to unleash a virus such as H5N1 bird flu in an effort to kill large numbers of people.

‘They can build drones, they can build robots, they can control drones, they can control robots, so they can spread this virus all over the place,’ he said, outlining the theoretical danger.

Concern over AI’s potential dangers has intensified around the world, roughly 70 years after the term was first coined. In recent months, industry figures and researchers have issued increasingly stark warnings about what could happen if super-intelligent AI develops without effective controls.

Evan Hubinger, an AI safety researcher at Anthropic, has estimated that there is a 10 percent chance AI could wipe out humanity within the next decade. Jacob Coxon, who has worked at both OpenAI and Anthropic, accused the companies of ‘gambling with our lives’ as they compete to build increasingly powerful systems. Anthropic CEO Dario Amodei, meanwhile, said a swarm of AI agents could seize control of the internet within six to 12 months.

Although many people have only recently begun considering these risks, some have spent years preparing for an extreme AI-driven disaster. Their plans range from fortified shelters and food supplies to equipment designed to protect against biological threats.

Some of the people building the technology are preparing as well. OpenAI co-founder Ilya Sutskever once said the company would need to ‘build a bunker’ before releasing AGI. Several technology billionaires have also invested in remote underground compounds, including Meta CEO Mark Zuckerberg, who is building a $100 million estate in Hawaii.

For people without billionaire budgets, Miller runs Fortitude Ranch, a network of eight survival sites in remote parts of the United States. Members can buy access to accommodation where they could ‘bug out’ during a societal collapse. A single person can secure a bed in a shared, basic room with a $1,947 down payment, followed by quarterly and annual fees. A large family seeking a private luxury en-suite room would pay at least $26,000 upfront.

The ranches grow their own food and maintain supplies and weapons intended to defend against both AI-controlled drones and human looters. They are designed to operate independently and are located away from power plants and data centers, which Miller believes could become targets.

‘If you’re near an AI data center or a power plant or a transmission line, that’s something that, in an AGI takeover, they will want to protect and they will not want pesky humans anywhere near that, so you’re likely to get killed,’ he said.

Fortitude Ranch was established in 2014 to prepare for a range of disasters, including civil war, nuclear attack and widespread power outages. Miller now says he is receiving a surge of emails from people concerned about the AI threat.

Miller is not the only preparedness specialist whose services have suddenly attracted greater interest.

Stockholm-based entrepreneur and mechanical engineer Ulrik Horn has developed a portable clean-air capsule that he says can shield occupants from biological agents in the surrounding atmosphere. Shipped on pallets, the tent-like structure can be assembled when needed and uses advanced air purification technology to protect against a biological weapons release.

Horn’s Bubble Boy-style Fonix capsules cost $50,000 each. He says five have already been pre-ordered, with AI safety researchers among the first customers to reserve one.

But Horn isn’t just getting emails about the Fonix tents. In the past week, friends and acquaintances have been coming to him with a range of questions on preparing for any hostile technological takeover.

‘One of my best friends just wrote to me, he was like “I have money in my bank account: What if a bot just sets us to zero?” And I was like, “Yeah, that’s a possibility,”‘ he said.

So Horn is starting an advisory firm to help people put measures in place to prepare for various hostile AI scenarios.

Another person deploying innovative tech to counter potential AI-instigated disruptions is Chris Sherwood, a tech YouTuber who has developed Project NOMAD, a downloadable electronic knowledge base which operates without the internet.

It can be loaded up with a host of resources, including maps, Wikipedia, survival manuals and even a whole school curriculum, said Sherwood, pointing to Amodei’s recent comments.

‘Having something like this available in that situation, when everyone is isolated, even if it’s only for a temporary amount of time, that’s where this really becomes super useful,’ Sherwood told the Daily Mail.

While it can be downloaded for free onto an existing computer, Project NOMAD’s website also has a suggested hardware set up complete with carry case, server, screen and portable solar energy source, which costs $2,400 in total.

Sherwood estimates that hundreds of thousands of people have downloaded Project NOMAD since he released it in March, and it has sparked lively debate on his social media channels.

‘This one just exploded right out of the gate, so it’s interesting to see that there’s definitely something that is striking a chord with people,’ he said.

While Horn and Sherwood are deploying cutting edge technology to keep the AI threat at bay, others are going for a more back to basics approach.

Cody Beckner is a member of the Virginia Kekoas, a militia whose members train in firearms, military tactics and survival prepping. They do regular outdoor survival operations, but Beckner is feeling overwhelmed by the potential AI threat.

‘AI poses so many different risks because the bad scenario could take on multiple forms at once,’ he said, adding that this made it difficult to prepare for.

‘If it’s a bioweapon, you’re just kind of screwed unless you have a bunker and you’ve got copious amounts of internally supplied oxygen with a gas mask and the proper filters and all that stuff, which most people just can’t afford,’ he said. Luxury bunker company Survival Condo, for example, offers penthouse units deep underground for $4 million.

‘The only thing I can think of is to just buy some land and go off-grid and minimize technological involvement.’

He has started growing his own food and learning to freeze-dry his supplies: ‘Realistically, the best plan you can have is to return to monkey: Try to be ready and able to drop out of society at a moment’s notice.’

Miller has similar advice for anyone wanting to evade rogue AI: Make yourself as unthreatening as possible.

‘A determined AGI system could kill all humans, but as a rational superintelligence, they should reason that humans living far from assets they use and value, posing no threat, and costly to exterminate, are not worth the resources to exterminate,’ he said.

Ulrik Horn is trying to stay optimistic and hopes that innovations like his biosafety tent might grab enough attention to spur policymakers to take action to put guardrails on the development of artificial intelligence.

‘Am I awareness-raising by showing a possible future, or am I preparing for it, or both? I don’t know anymore,’ he mused.

‘Let’s just come together and let’s not take unnecessary risks, it’s in nobody’s interest. What I would like the most is to not have to do this job.’