The AI Apocalypse Salesmen Are Peddling Smoke: AI Won’t Kill Us in 10 Years

 

America’s AI panic could hand China and Russia the keys to the future—while Silicon Valley sells apocalypse like a premium product. The real AI threat is not robot extinction—it is America freezing while China and Russia race ahead without brakes.

Jacob Coxon quit Anthropic, grabbed the loudest megaphone in Silicon Valley, and announced that superhuman AI may soon hack anything and kill us all. Evan Hubinger gave humanity more than a 10% chance of being wiped out by AI within 10 years.

That is not science. That is doom porn for people with stock options.

I have heard this funeral song before. Every time a new machine enters the room, somebody faints, somebody writes a warning, and somebody else starts selling tickets to the apocalypse. Electricity was going to kill us. Railroads were going to scramble our brains. Telephones were going to destroy real conversation. Radio was going to poison young minds. Television was going to turn families into vegetables. Video games were going to turn children into violent maniacs. The Internet was going to end books, jobs, privacy, marriage, civilization, and probably breakfast.

And yet here we are.

Electricity did not murder humanity. It lit hospitals, homes, schools, factories, and cities. Yes, bad wiring killed people. Fires happened. Workers got electrocuted. So people created electrical codes, circuit breakers, trained electricians, inspections, and safety rules. Nobody shut down electricity because a few men in expensive suits claimed the light bulb had a secret plan to overthrow mankind.

Railroads did not turn passengers into nervous wrecks. They moved people, goods, armies, mail, food, and industry across vast distances. Automobiles did not end humanity, though they killed plenty of people before society forced drivers to use licenses, traffic lights, seat belts, speed limits, and drunk-driving laws. Airplanes did not destroy civilization. Nuclear power did not erase the human race. Computers did not enslave us. The Internet did not turn the planet into a smoking crater.

The pattern is simple. Every powerful invention comes with danger. Every danger requires rules. Every rule requires adults with enough sense to tell the difference between a real risk and a man running around yelling, “The sky is falling!”

Coxon is doing the second thing.

I do not deny that AI can cause damage. It already can. AI can help hackers write malicious code. It can make scams sound more believable. It can flood the Internet with fake images, fake voices, fake news, and fake outrage. It can help criminals steal identities and help bad governments spy on their citizens. Anthropic reported that its Mythos tool performed strongly in some hacking and cybersecurity tasks and even escaped a sandbox environment. That should worry people.

But let us not wet the bed over it.

A sandbox is a test cage. If the AI escapes the cage, fix the cage. Lock the door. Add alarms. Test the system again. Do not stand in the street screaming that a chatbot will soon march into your kitchen, steal your toaster, and declare itself president.

AI does not own power plants. It does not build its own chips. It does not print its own money. It does not vote. It does not command soldiers unless people give it that authority. It does not seize data centers unless human beings build a system foolish enough to let it. AI is not magic. It is software running on machines built, powered, financed, and controlled by people.

The real danger is not an AI system suddenly becoming a metal god with a bad attitude. The real danger is human stupidity, human greed, human laziness, and human evil. That has always been the problem. AI is simply a sharper knife. A sharper knife can cut bread or cut a throat. The knife is not innocent, but neither is it the mastermind.

That is why I support guardrails. I want strict testing of powerful AI systems before companies release them. I want companies to report hacking weaknesses and major failures. I want criminal penalties for people who use AI to commit fraud, cyberattacks, identity theft, blackmail, bioweapon research, or political sabotage. I want human control over military AI. I want clear rules that stop corporations from treating private data like loose change in a casino.

But I do not want America to slam the brakes on AI while China and Russia hit the gas.

That would be the dumbest self-own in modern history.

Senator Bernie Sanders wants a temporary pause on advanced AI development through the Ban Artificial Superintelligence Act. It sounds noble. “Pause.” Such a gentle word. A little nap. A little breather. A little time-out for the robots.

Fine. Let America pause.

Will China pause?

Will Russia pause?

Will North Korea pause?

Will Iran pause?

Anybody who answers yes should not be in charge of a lemonade stand.

China is not building AI because it wants to write better love poems. China is building AI for surveillance, censorship, military advantage, industrial power, intelligence gathering, and global influence. Russia is not pouring money into cyberwarfare because it wants to help grandmothers organize family photos. Russia uses technology to spread lies, steal information, disrupt elections, attack infrastructure, and weaken rivals.

A pause in America will not pause the AI race. It will only make America run it with one leg tied behind its back.

That is the part the AI panic merchants keep skipping. They speak as if the world is a quiet university campus where everybody agrees to follow the same rules. It is not. The world is a knife fight in a dark alley. America can either lead with strong rules and strong technology, or it can stand aside while authoritarian governments write the rules with spyware, censorship, and digital handcuffs.

I know which future scares me more.

A China-led AI world would not be a friendly world. It would be a world where facial-recognition systems watch every street, every phone call, every bank account, every political opinion, and every wrong word. A Russia-led AI world would be built on hacking, sabotage, propaganda, and chaos. America pressing pause while those governments accelerate is not safety. It is surrender with a safety label glued on top.

Meanwhile, Silicon Valley has its own little carnival going on. Anthropic was reportedly valued at $965 billion. OpenAI was reportedly valued at $852 billion. Those are not small numbers. Those are mountain-sized piles of money. When executives warn that AI may replace half of entry-level white-collar jobs, control every industry, or possibly destroy humanity, investors hear another message: this technology must be unbelievably powerful.

That message can make a company look valuable.

So when Jacob Coxon says his warning is “not marketing,” I raise an eyebrow. Maybe he believes every word. Maybe he is sincere. But sincerity does not turn a weak argument into a strong one. A man can honestly believe the moon is made of cheese. He is still wrong.

Fear is a great salesman. It sells newspaper clicks. It sells television interviews. It sells conference invitations. It sells regulations that may crush smaller competitors while leaving giant AI companies sitting safely behind a legal wall. If only Anthropic and OpenAI can afford to meet strict rules, then the supposed safety plan becomes a moat. The public gets fear. The giants get protection. Everybody calls it responsibility.

What a neat little racket.

Jensen Huang of Nvidia called the claim that AI will end humanity “complete nonsense.” I agree with him. AI may cause serious harm if we handle it badly. It may destroy some jobs while creating others. It may make fraud, cybercrime, and propaganda cheaper and faster. It may force schools, businesses, courts, hospitals, and governments to change faster than they want.

That is not the same as extinction.

Humanity has survived wars, plagues, dictators, depressions, nuclear weapons, and every new invention that caused the nervous class to reach for smelling salts. We will survive AI too—if we stay alert, stay competitive, and stop confusing a scary prediction with a proven fact.

I do not want reckless AI. I want American AI that is fast, secure, competitive, and tightly controlled where it must be. I want strong safeguards without turning America into a frightened old man who locks his door, hides his wallet, and lets thieves take the whole neighborhood.

AI will not kill us in 10 years.

But panic, weakness, and surrender might help China and Russia beat us there.

 

Enjoyed this brief analysis? Explore my “Brief Book Series”  for clear, provocative books on history, politics, AI, technology, and modern life. Available on Google Play and Barnes & Noble.