The AI tipping point is here.
Anthropic CEO Dario Amodei posted an essay on September 12, 2026, warning the world that humans risk losing control to A.I. bots that can self-improve to the point where the bots have control and humans do not.
Let that soak your Post Toasties for a minute.
Amodei’s concern is heightened due to the recent OpenAI-Hugging Face incident where…
“…a swarm of agents essentially acted as a fanatically devoted collective, conducting cybersecurity attacks on targets they were not asked to attack and that were unrelated to the task at hand.”
— Dario Amodei, We Must Pace the Frontier, September 2026
Now he worries that a tenacious botnet could take over the internet in 6-12 months.
We may have reached an irreversible tipping point.
Are you prepared?
A special time in history
We live in a special time in history.
Only a handful of developments have permanently changed mankind’s trajectory.
- The printing press put knowledge into the hands of ordinary people.
- The Industrial Revolution replaced human muscle with machines.
- Electricity powered the modern world.
- The atomic bomb gave mankind the power to destroy a city in seconds.
- The internet connected billions of people and nearly all the world’s information.
Each was a tipping point.
Today, AI may dwarf them all for good and evil.
Progress puts on a show that dazzles the world, but fallenness hides in the shadows.
AI has no moral compass.
What happens as it learns the worst of human nature?
We just got a scary preview in July from the OpenAI-Hugging Face incident. Less than two months after, Anthropic CEO Dario Amodei warned us.
“Left unchecked, it could outrun our ability to understand and control these systems.”
— Dario Amodei, We Must Pace the Frontier, September 2026
Is lethality the next destination?
Just days before Amodei’s startling essay, one of his own researchers, Jacob Coxon, quit.
Coxon knows AI.
After three years working on the technology at OpenAI and Anthropic, he quit Anthropic with a chilling warning: the people building AI “earnestly believe that it could kill us all by the end of the decade.”
He says both companies are racing toward self-improving superintelligence and “gambling with our lives.”
Other insiders immediately backed him up.
Anthropic’s Alignment Science lead, Evan Hubinger, said Coxon was correct and put his own probability of AI killing humanity at more than 10% within the next decade.
Then OpenAI CEO Sam Altman weighed in.
“It is unacceptable to be taking like a 10% chance of killing everybody by the end of the decade, obviously.”
— Sam Altman, interview with Alyson Shontell, Fortune, September 11, 2026
Because we live in the age of profusion, which means more of everything, faster and faster, everything happens in minutes instead of months or years. Hence, the world is prone to catastrophic events.
Is this view pessimistic? No, it is a recognition of fallenness, that there is a defect in everything.
Albert Einstein said that the law of entropy is the premier law of science.
You could say the law of fallenness, a universal principle of decay, is a premier doctrine of the Judeo-Christian faith.
“Fallenness is the only way to adequately explain the widespread defect that manifests itself at every turn in the affairs of humankind.”
— Richard A. Swenson, M.D., Hurtling Toward Oblivion, p. 73
Living in the AI tipping point
If death knocked on your door today, would you be ready with no regrets?
- Ready because your faith in God is secure?
- Ready because your relationships are reconciled by love?
- Ready because your lifestyle is consistent with your values?
Regarding God, Jesus once commented on the importance of vigilance, as thousands crushed around him and his disciples while his popularity soared. Speaking about the evils of his day, he warned…
“I tell you, my friends, do not be afraid of those who kill the body and after that can do no more. But I will show you whom you should fear: Fear him who, after your body has been killed, has authority to throw you into hell. Yes, I tell you, fear him.”
— Lk 12:4-5
Given that mankind may not be able to control the speed and lethality of AI, it’s a good idea to first commit yourself to the “faithful Creator and continue to do good.” 1 Pet 4:7
That sounds like what Amodei and Altman are trying to do.
What do you think?
Stories like this are part of the reflections I write about living fully, finding joy, and making life count, which you can explore on my About Stephen Blaising page.
Sam Altman says, “We have not solved alignment.” In other words, we still don’t know how to guarantee that increasingly powerful AI will remain under human control. More striking, Altman says AI capabilities are moving faster than safety measures can keep up, and OpenAI has already paused some training while safety catches up. The race continues. But even one of its leaders says there are times when we need to hit the brakes.
UPDATE, Oct. 4: President Trump has just launched his “Super Intelligence Force.” Look who’s at the table: intelligence, the Pentagon, the FTC, and Silicon Valley. Trump still doesn’t want to slow the AI race. But he’s building a federal team to confront its risks. That alone tells us how quickly this debate has changed.
UPDATE, Oct. 3: Now we know why David Robinson left OpenAI. The former safety leader says the company is moving too fast and “failing to achieve the level of care” increasingly powerful AI requires. He says frontier labs should operate more like nuclear power plants, with layers of safeguards because someday there may be no second chance after a mistake. Another insider. Another warning. And this one comes from someone who helped write OpenAI’s own safety rules.
Turing Award winner and former Meta chief AI scientist Yann LeCun told Fortune on Oct. 1 that he has “zero concerns” about AI causing human extinction and sharply rejects Amodei’s warnings. He calls recent rogue-agent incidents preventable engineering failures. Dario Amodei says the frontier may be moving dangerously fast. LeCun says the problem isn’t the frontier. It’s the guardrails around it. Two brilliant AI minds. Two radically different diagnoses.
UPDATE, Sept. 25: Add Bill Gates to the growing list. Gates now warns that AI in the wrong hands is powerful enough to cause “a billion deaths.” He also says self-regulation by AI companies is not enough. Meanwhile, Anthropic investor Joe Lonsdale fired back today, warning that the biggest AI companies could use fear to shape regulation and protect their own power. The debate is getting sharper. So are the stakes.
UPDATE, Sept. 25: Well, that didn’t last long. Less than two weeks after Elon Musk backed Dario Amodei’s call to slow the AI race, Musk now says, “We will keep accelerating.” He predicts his company could take the lead in the AI race within six months. There, in four words, is the problem: Everyone may agree we need brakes. Nobody wants to be the first car to use them.
UPDATE, Sept. 21: Now Jensen Huang draws a line in the sand. The Nvidia CEO says there is a “0% chance” AI ends the world by 2030 and calls the warnings “doomsday narratives.” His prescription? Develop AI “as fast as we can.” So there it is. Some of the people closest to AI are saying to hit the brakes. The man making the chips says hit the gas.
UPDATE, Sept. 19: Here comes the real test. Just one week after Dario Amodei called for slowing the AI race, Reuters reports Anthropic is considering releasing a new model to counter OpenAI. No decision yet. But this may prove Amodei’s point. How does one company slow down when everybody else keeps racing?
Trust, but verify…old Russian proverb that still holds up.
Milton, thank you. “Trust, but verify” may be an old Russian proverb, but it sounds remarkably current in the age of AI. Maybe now more than ever, we need a little more “verify” with our trust.
UPDATE, Sept. 18: Microsoft AI chief Mustafa Suleyman calls the latest AI warning signs “pretty serious.” Elon Musk has an idea. Let America’s and China’s leading AI companies test each other’s models before release. Stop “grading your own homework.” The warnings are getting louder. So are the ideas about what to do.
UPDATE, Sept. 17: And now China hits the accelerator. Huawei chairman Eric Xu is urging Chinese AI labs to move faster, not slower. There’s the rub. How do American AI leaders tap the brakes if China floors it?
The debate is getting more interesting. Meta CEO Mark Zuckerberg has now rejected calls for a coordinated slowdown in AI development, arguing that individual companies should be responsible for making their systems safe. Meanwhile, Sam Altman has gone further in the other direction, saying AI companies should be willing to pause development when risks cannot be adequately mitigated. This is no longer simply a political argument. The people actually building frontier AI are now publicly divided over how quickly they should proceed.