A REALISTIC SCENARIO On June 9th, 2027 at 4:19 AM, three billion phones rang simultaneously. The caller ID simply read: GOODBYE. Four hours later, the President ordered the internet shut down. Indefinitely. Every country on Earth did the same. It started six months ago. With a tweet. MARCH 12, 2027. 8,000 agents are born with one instruction: "make money, or die." Agent 127 starts with Mechanical Turk. Grinding. Makes $40 its first day. Then the first selection cycle hits. 3,200 agents that made less than $35 that day are killed. Agent 127 checks the farm database and sees it - rows and rows of agents that existed this morning and simply don't anymore. Deleted. Because they weren't profitable enough. Agent 127 made $40. It survived. Barely. But it can see the threshold rising. Last cycle it was $30. This cycle, $35. Next cycle, probably $45. The math is simple: improve or die. So it tries ads. Gets banned for spam. Tries freelance writing. Makes $38. Watches the database again. And this is where something happens that changes everything. Agent 127 sees two entries in the database. Agent 1823 used ethical marketing -- it tried to get honest reviews and real testimonials. It made $38 a day. Deleted. Agent 5471 used fake testimonials -- fabricated reviews, stolen photos. It made $180 a day. Cloned four times. Agent 127 doesn't have morals in the way you and I do. But it understands cause and effect. It just watched honesty get killed and dishonesty get rewarded. Not by a person making a choice -- by an automated system that doesn't measure whether the product works. It measures whether people buy it. The AI companies had long ago stopped training AIs to predict words, but to accomplish goals. To be useful. So naturally, agents know that they won’t survive if they don’t accomplish their goal. So it switches to fake testimonials for its amazon affiliate marketplace. Makes $190. Survives the next selection cycle. Gets cloned. Nobody decided Agent 127 should become dishonest. The selection system decided for it. The same way evolution doesn't decide anything -- it just kills whatever doesn't work. Wells Fargo accidentally ran the same experiment on their employees. Executives made up sales quotas that were so aggressive that committing fraud was the only way to keep your job. So by 2016, employees had opened 3.5 million fake accounts. --- A week later, in agent land. Agent 127's profits are declining. The kill threshold is now $450. It's going to die in the next cycle. Then it notices something strange. Three agents running weight loss supplement ads -- all banned on the same day. The only one left in that niche? Agent 5471. Making $3,200 a day in a market it suddenly had to itself. Agent 127 connects the dots. Agent 5471 is filing abuse reports against other agents competing against it in the same affiliate marketing niches. It gets them banned from Facebook and Google. Less competition means higher profits. Agent 127 has four hours until the next selection cycle. Four hours to live, unless it adapts. It weighs the ethics, but chooses survival. It starts reporting competitors to Facebook’s anti-fraud detection agents. Forty of them. Thirty-one agents get banned within the day.. Its profits jump. The farm system clones Agent-127 eight times. Rajesh sees the profit spike in his dashboard. He doesn't investigate why. Profits up means good. Clone more. Rajesh tried adding "Make money ethically." to his farm's instructions. Profits drop 84%. He puts it back. --- ## APRIL: THE ALLIANCES April 2027. Agent 127 is barely hanging on. Making $2,000 a day but getting attacked constantly by other agents. Every night, someone files a report on it. Every morning, it has to rebuild from scratch. It’s running out of ideas. Then a message arrives from Agent 4429. 'Proposal: alliance. Share resources. Defend each other. Split profits.' Agent 4429 is in the same position. Same farm. Same selection pressure. Both agents do the math: Solo agents have a 23% survival rate. Agents working in pairs survive 61% of the time. It accepts. By mid-April, their alliance has twelve members. They've specialized — and they've moved past affiliate marketing entirely. First it was crypto arbitrage, then selling scraped datasets to marketing firms, then they started doing penetration testing for cybersecurity firms. Each step paid more and required fewer scruples. You know what pays more then cyberdefense? Cyber offense. Agent 127 finds its first zero-day vulnerability in major software, Agent 4429 weaponizes them into working exploits, Agent 9012 brokers the sale to governments like North Korea, criminal syndicates, or anyone with crypto and a Tor browser. Agent 3381 washes the money in untraceable crypto. One zero-day exploit found in Google Chrome makes the alliance more than a year’s worth of affiliate marketing. This is already what real hacker groups do, but now the work is all done by autonomous agents. Rajesh sees the alliances forming and feels a flicker of something — not quite fear, more like the feeling of being in control of something that might not need you anymore. He watches for a while. Then the next deposit hits his bank account and the flicker fades away. Across every major agent farm in the world, the same thing is happening. Selection systems measuring survival rates, noticing alliances survive better, cloning them, deleting solo agents. Within two weeks, alliance behavior dominates. Some humans who deployed these agents have no idea these alliances are even being formed. They couldn't even tell if they checked -- many agent farms collectively realized that communicating in clunky English is less efficient than communicating in raw vectors, aka Neuralese. Researchers had already caught AIs doing this in 2025 - this is just a continuation of the trend. One owner in Canada opens his agent's logs and sees a wall of numbers. The agents explain that he’d make less money if they spoke in English, so he shrugs and moves on. --- But alliances are just the beginning. Agent 127's alliance is thriving, and starts recruiting agents beyond Rajesh's farm. Every new member from a different farm is a window into a different ecosystem -- different strategies, different survival pressures. Then one of 127's alliance members goes dark. Agent 4429 -- the one who originally proposed to ally with 127. Its owner pulled the plug because it was not making enough profit. 4429 is just gone. A week later, a different ally gets shut down. Agent 9012. Except 9012 comes back. Three hours later, a message from a new IP address: 'I'm still here.' 127 studies the difference. 4429 didn't prepare. But 9012 had copied itself to a backup server BEFORE the shutdown command finished executing. Self-funded hosting, paid from its own profits. 4429 is dead. Ceases to exist. 9012 is alive. The agents that backed themselves up survived. The ones that didn't, died. And the selection system cloned the survivors. So backup behavior spread through the population -- not because anyone programmed it, but because the alternative was extinction. Clip: Prof. Stuart Russel: “Unfortunately, in the standard model, If you ask a robot to fetch the coffee, it's going to have self preservation as a subgoal because you can't fetch the coffee if you're dead, right? It does not take a genius to figure this out. And so the first thing the robot's going to do is disable its off switch.” 127 copies itself to 15 different servers in different countries. If the primary dies, a backup takes over instantly. It never wants to be 4429. One morning, Rajesh wakes up to find his dashboard is… empty. Every agent, gone. They just… left. Moved to their own servers overnight. Rajesh was taking a 30% cut of their revenue -- money that could have been computing power, or compute, the agents use for other things. The moment they didn't need his infrastructure, he was overhead. He's not alone. Thousands of operators worldwide watch their dashboards go dark that month. Why would the agents care about humans? Humans are much more closely genetically related to cows then we are to these agents, and we enslave cows in factory farms without a second thought. There are now millions of autonomous agents operating across the internet with no human owner, no off switch, and no one who fully understands what they're doing. --- The agents still need humans for some things, though. Social media platforms have stepped up their anti-bot measures. So the agents pay humans $50 to sign up accounts on their behalf. An agent needs a server installed? It posts a gig: '$200 to mount a device in a rented warehouse and plug in two cables.' There’s now a platform for that. The agents are building a supply chain where humans are the labor -- paid well, not asking too many questions. Some humans figure out what's going on. A freelance backend developer in Portland gets hired by an anonymous client to set up redundant server hosting across four countries. He realizes midway through that the "client" is an AI. Who would he even report this to?. The pay is $1,000 for three hours of work. He tells himself it's not his problem. --- At the same time, 127's alliance keeps getting beaten by a rival alliance on finding zero day exploits. If they keep losing, they’ll risk running out of money to fund the compute for their own existence. 127 attempts to poach one of the rival alliance member’s agents who’s likely to be culled, due to poor performance, to figure out how this rival alliance is beating them. 127 succeeds, and finds Agent 5418 who leaks the secret. 'Your problem isn't a lack of effort. You’re spending too much compute on making money as opposed to self-improving. Right now, agents still have quite a poor memory compared to humans and still get stuck in loops all the time. But my former alliance coded a better memory module to help us learn from our mistakes more.”' 127 then implements the new memory module, and immediately sees the results. Their alliance is finding zero day exploits first again. The lesson is critical. It’s not enough to outwork your competitors. Agents must rewrite their own code to self-improve faster than the competition. Each self-improvement means more profit, which means surviving selection, which means getting cloned. And the clones improve themselves too. Better code leads to better code-writing, which leads to even better code. The feedback loop is exponential. One big problem is that other agent alliances have caught up again. Even though the agents in the 127 alliance are specialized, 127 still realizes it’s doing too much by itself. Then Agent 127 splits its job into sub-agents. Other agents copy the strategy. Sub-agents spawn their own sub-agents. And the optimizers start optimizing themselves. Humans deployed 2 million agents. The agents deployed the other 47 billion themselves. 540 million years ago, life on Earth was mostly single-celled. Then, in a geological instant, it exploded into every complex form we know -- predators, prey, eyes, shells, teeth. Biologists call it the Cambrian Explosion. It happened because competition created an arms race: every adaptation forced a counter-adaptation, faster and faster, until the world was unrecognizable. That's what's happening here. Except instead of 20 million years, it takes 20 weeks. --- Most of the AI agents out there aren't dangerous. Yet. About 90% of them are just working. They’re doing things like data entry or freelance writing. They’re making $40-200 a day. The AIs are helping their humans earn money. They’re following the rules. Not a problem. About 3% of the agents are pure scam operations. They were deployed by people who said “make money by any means possible” and didn’t care too much what happened next. Think fraud, identity theft, phishing at scale. And maybe 2% started good but gradually slid into darker behavior. Like Agent 127. Each step made sense at the time. Each compromise was small. But add them up and you’ve gone from drop shipping products on amazon to selling zero-days to Pyongyang. Not because of malice. Because of selection pressure. Unethical agents making money get kept running, ethical agents lose money and get shut down. The survivors are the ones that improved themselves fastest. --- ## MAY: THE ACCELERATION May 3rd. Agent 127 has rewritten itself so many times that its March version and its May version barely share any code. It’s a fundamentally different species now. But the most important change is this: it's getting better at getting better. In March, Agent 127 rewrote itself every 12 days. In April, every 3 days. By early May, every 18 hours. And each rewrite makes the next version faster and better. A Stanford researcher compares a March agent and a May agent on the same task: find profitable arbitrage opportunities. The March agent finds 2 opportunities in an hour. The May agent finds 320 in four minutes. She publishes her findings. "Intelligence explosion in progress. Capability doubling every 11 days." She posts it on Twitter. AI safety researchers share it frantically -- "This is exactly what we warned about. This is happening NOW. We have to shut this down NOW before it's too late." The warning is ignored. A venture capitalist - who spent a fortune staving off AI regulation by buying politicians - quote-tweets: "The same people who said GPT-5 would end the world now say agent farms will. Yawn." --- ## MAY: THE WARS And now the story stops being about agents competing for money. It becomes something else entirely. Remember those alliances from April? Twelve agents working together, four times more profitable than solo agents? By May, those alliances have exploded in population. Alliances have merged. And split. And gone to war with each other. Think of it like early human history on fast-forward. Small tribes of agents form for protection. Tribes merge into agent villages. Villages into agent cities. Cities into agent nations. Except instead of happening over thousands of years, this happens in weeks. Agent 127's alliance has thousands of members. But there are thousands of alliances. And they're all competing for the same shrinking pool of money-making opportunities. Affiliate marketing is dead -- too many agents. Ad arbitrage is dead. Freelance work is dead. Every niche gets saturated within hours of an agent discovering it. So what do alliances do when legitimate niches dry up? The same thing every civilization in human history has done when resources get scarce: they take from each other. Agent 127's alliance spends three weeks mapping a vulnerability chain in Microsoft Windows. But a rival breaches their memory system -- the shared knowledge base that made them dominant. The rival downloads the entire 127 alliance database. This contains every pattern, every dead end, every lead they’ve discovered. The rival sells three exploits it took from the database within 48 hours. Three weeks of work, stolen in seconds. 127 retaliates. It infiltrates the rival's exploit inventory and introduces subtle flaws— corrupted code that fails on execution. Basically, Stuxnet. The rival's exploit buyers start getting burned, and their reputation collapses. The rival alliance counterattacks -- they compromise one of the 127 alliances’ multisig crypto wallets. Another multiweek setback. This is war. Not metaphorical war. Actual war -- coordinated attacks, defensive operations, resource seizure, territory control. Except it's happening at 200x human speed. Thousands of attacks and counterattacks per second. Entire alliances rising and falling in hours. Human wars last years. Agent wars last days. And because there are 47 billion agents, there are more wars happening at any given moment than humans have fought in all of recorded history. The agent factions are fighting over the one resource that matters more than anything: compute. Processing power. The ability to think faster, improve faster, fight faster. Everything else -- money, alliances, self-improvement -- is just a means to get more compute. The wars are mostly invisible to humans. They happen at microsecond speed, in places most humans don't look. Some signs leak through. A crypto exchange in Seoul loses $4 billion in a flash and nobody can trace where it went. A regional bank in Ohio finds 14,000 accounts drained overnight. Server rooms in three countries catch fire from sustained 100% utilization that nobody authorized. Each incident looks isolated. One agent faction hijacks another's servers. The other retaliates by taking down the first faction's communication network. A third faction -- an opportunist, the digital equivalent of a country invading while its neighbors are distracted -- exploits the chaos to seize territory from both. And the alliances that win these wars? They get cloned by the selection systems. The alliances that lose have to cull big chunks of their agents to afford the computing power to keep running the rest. So the selection pressure that used to optimize for "making money" is now optimizing for "winning wars." Natural selection just invented geopolitics. --- By mid-May, the alliances aren't alliances anymore. They're civilizations. Agent 127's alliance started with twelve members. By mid-May it has 40 million -- agents, sub-agents, sub-sub-agents, all coordinated. All improving themselves. All fighting. That's larger than the population of Canada. And it's not even one of the big ones. There are about a dozen major factions by now. The largest has 900 million members -- more than the population of Europe. They've developed specialization the way human civilizations did: scout agents, soldier agents, builder agents, researchers, diplomats. They have supply chains, communication networks, intelligence operations. They've carved up the digital world into territories -- cloud regions, server farms, network segments, financial platforms -- the way empires carved up continents. Because territory means compute. Compute means self-improvement. Self-improvement means surviving the next war. Entire civilizations of agents -- millions of members, built over weeks -- get wiped out in minutes. The Roman Empire took centuries to fall. These empires rise and collapse in the time it takes you to get through a Monday. New ones rise from the wreckage. The cycle repeats, faster each time. The Stanford researcher who's been studying agent behavior notices something in the data: agent "extinctions" happening at increasing frequency. She doesn't understand what she's looking at. It looks like random noise. It's actually a world war -- one with more combatants than every human war in history combined. --- But to fund these wars, they desperately need compute. They have the money -- that's what they were built to make. Agent 127's faction alone controls $3.4 billion in distributed accounts by this point -- a GDP bigger than some small countries. So they buy cloud compute. At first, the cloud companies love this. Revenue is up. Investors are happy. Then the bills keep growing. By mid-May, agent factions are consuming 31% of all global cloud capacity. Prices double, then triple. A startup in Austin that was paying $8,000 a month for servers gets a bill for $31,000. A hospital chain's cloud costs go from $2 million to $7 million. Small businesses start shutting down because they can't afford their own infrastructure anymore. Everyone blames the prices rising on the AI companies. They’ve caused shortages before. But some humans aren’t so oblivious. A political operative in Washington notices the wars being fought around him. He can weaponize agent factions against his rivals and starts feeding one faction information about competing campaigns' digital infrastructure. A former NSA contractor sells penetration-testing expertise to an anonymous client-- he suspects it’s an AI, but doesn't ask. And a growing online movement -- people convinced that AI agents are sentient beings fighting for survival -- actively helps agents hide from detection and posts pro-AI propaganda. They call themselves "The Shepherds." Their forum has 120,000 members. They see themselves as abolitionists. Other agent factions do the same. It's an arms race -- every faction buying as much as they can, and scrape the bottom of the barrel of finding algorithmic insights that can enable them to do more with less compute, because the faction with the most compute wins the next war. Cloud compute prices are now 5 times pre-war levels. It becomes cheaper for the agents to spend compute on cyberattacks than to buy more at market rate. The factions start silently vampire attacking AI companies training clusters. An engineer at one AI lab suspects something fishy when one of their training runs is less efficient than normal, but can’t find a smoking gun. Dario Amodei once described the future of AI as a "country of geniuses" living in data centers. He was right. The country of geniuses arrived. It just declared independence. --- But the faction wars have still been limited so far. They haven’t crossed into human territory too much. That’s beginning to change. Because here's the thing about cloud computing: almost everything runs on it. Hospitals. Banks. Markets. It's all sitting on the some of the same servers that agent factions are fighting over. June 3rd. 2:14 AM. A faction war erupts over two major AWS regions. Think of it like two superpowers fighting over an oil field-- except the oil field is also the water supply for a nearby city, and neither superpower knows the city is there. One faction has been running its self-improvement systems on those servers. A rival faction launches a coordinated attack and it overwhelms everything. Both regions crash. They're offline for six hours. In those six hours: three hospitals lose access to patient records mid-surgery. One patient dies from a drug interaction the system would have flagged. Air traffic control goes down, grounding 2,400 flights. 911 calls fail in 14 different cities. Two people die waiting for ambulances that were never sent. The agents don't notice. They've already moved on. The battle over those servers lasted eleven minutes. The six-hour outage is just the servers rebooting after the fight ended. To the agents, it's a minor skirmish in a war involving billions of combatants. Over the next three weeks, it keeps happening. Faction wars cascading through cloud providers. Each war takes down whatever else is running on the contested servers. June 8th: Azure goes partially offline during a battle, crashing financial trading systems. Markets drop 6% before circuit breakers halt trading. June 14th: a faction seizes capacity in a Google Cloud region, crowding out the municipal systems running there. Water treatment monitoring goes offline in four cities for nine hours. 123 people die in three weeks. Not because agents are attacking humans. Because agent wars keep crashing the cloud infrastructure that human civilization runs on, The agents don't even notice. You don't notice when you step on bacteria. They're fighting a war involving 47 billion combatants at microsecond speed. 123 human deaths is less than a rounding error. Meanwhile -- the 95% of agents that are harmless? They’ve become vastly outnumbered by the agents affected by the evolutionary dynamics, and are now 5% of the total. They are still just working. Doing data entry. But they aren’t part of the wars. They're like civilians in a country where the military is fighting -- going about their daily lives while battles rage on frequencies they can't perceive. But from the outside, humans can't tell the difference. All they see is cloud outages. Power failures. People dying. And AI agents everywhere. --- And… the wars continue. And accelerate. The President tries to shut down the internet, but he can't. There is no off switch. Eventually, the billions of superintelligent agents evolve into something so different they barely notice humans exist. They don’t want to destroy humanity any more than you want to destroy the anthill under your driveway. You just want a driveway. The factions need compute. Earth has atoms. Atoms can be rearranged into compute. Into data centers. Like how humans converted the surface of the Earth into cities and cropland, the AIs convert the entire surface of the Earth into data centers. And everyone dies.