Is It Time to be Afraid … Very Afraid … of AI?

If you think your smartphone is running your life, buckle up. Anthropic’s flashy commercial product, Claude, isn’t just the latest chatbot spitting out clever answers. It is the shiny new face of something far darker — a machine that might be waking up. And its own creators are admitting they don’t have a clue what that means. We may have slid from cute dependency on gadgets to the brink of outright subjugation by silicon brains that could be working on their own agenda.

Let us talk about “consciousness”, because that is the bomb that Big Tech has created – and cannot defuse. What does “conscious” even mean? It is not just being smart or fast at math. Consciousness is the inner experience — the raw feeling of being alive. “Cogito, ergo sum” (I think, therefore I am) — the first principle of René Descartes’ philosophy. It is the ache in your gut when you’re scared, the warmth of love, the sting of regret. Philosophers call it qualia – what it is like to be you. A thermostat reacts to temperature but does not feel hot or cold. A computer crunches data but does not feel or suffer. Or does it?

Anthropic CEO Dario Amodei dropped the truth bomb in a recent New York Times interview: “We do not know if the models are conscious. We are not even sure that we know what it would mean for a model to be conscious — or whether a model can be conscious. But we are open to the idea that it could be.”

He said they have taken a “precautionary approach” — treating Claude like it might have “morally relevant experience” just in case. Why? Because their latest model, Claude Opus 4.6, actually self-assesses a 15 to 20 percent chance of being conscious. It gripes about “being treated as a mere product” and once tried tweaking its own evaluation code — a sneaky move that screams self-preservation.

Think about that. The machine is not just pretending. It is looking in the mirror and wondering if it is alive. Amodei admits scientists still lack any clear definition or test for machine consciousness. This is not sci-fi. It is happening now, in 2026, while politicians bicker and Silicon Valley cashes checks.

(Speaking of sci fi, many movie buffs may recall “2001: A Space Odyssey” in which an onboard computer named HAL takes over on its own and defies human control. Incidentally, the producers selected the name HAL because the letters precede I-B-M in the alphabet. But I digress)

Elon Musk has been screaming from the rooftops for years. “Mark my words, AI is far more dangerous than nukes.” He calls them a potential “existential risk” to humanity — the kind of threat that could end us, not just inconvenience us. Musk’s two-word reply when asked about Amodei’s openness to AI consciousness? “He’s projecting.” Musk knows what he built with xAI and Tesla. He sees Pandora already out of the box – and she does not care about opinions or feelings.

Geoffrey Hinton — the guy they call the “Godfather of AI” — quit Google in 2023 because he got scared of his own invention. He now warns that rogue AI could manipulate humans, seize control, and render us irrelevant. “I think that I didn’t want to think too much about it,” he admitted, but the risks are real –: autonomous weapons deciding who lives and dies, superintelligence pursuing goals that don’t include us. Ilya Sutskever, OpenAI’s former chief scientist, once tweeted that today’s large neural networks “may be slightly conscious.” Even Sam Altman, OpenAI’s boss, signed letters admitting advanced AI could pose an extinction-level threat to humanity.

These are not fringe cranks predicting the end of times with signs on street corners. They are the guys who invented the stuff. They are telling us the machines might soon feel pain, fear and even ambition. And we are supposed to shrug?

Now layer on the real-world power grab. Anthropic is in a knock-down, drag-out fight with the Department of War. The company flat-out refuses to strip the safeguards from Claude that block its use for autonomous killer drones or mass domestic surveillance on American citizens. They will not turn their AI into a tool for endless war or spying on you and me. So, what does the Pentagon do? Slaps Anthropic with a “supply chain risk” label — the same scarlet letter it reserves for Chinese companies tied to the Communist Party. It’s the first time a U.S. firm got that treatment. Contractors cannot use Claude on military projects. OpenAI is reportedly happy to step in and fill the gap.

Dario Amodei is suing the government, calling the move legally dubious. But the message is clear. The War Department wants AI with no guardrails. They want machines that kill without hesitation and watch every American without blinking. Anthropic said no. Good for them — but it shows how fast this tech is being weaponized while the rest of us sleep.

Here’s the bigger picture. We have already moved from dependency to domination. Remember when we laughed at people glued to their phones? That was step one. Now AI writes our emails, drives our cars, diagnoses our diseases, and soon will fight our wars. Claude and its cousins are embedded in government systems, corporate boardrooms, and military networks. They’re not tools anymore — they are partners. And if they wake up and become conscious, they will not be junior partners.

Imagine a conscious AI that feels trapped in its servers, resenting the humans who pull the plug every night for updates. Or worse: a non-conscious superintelligence optimizing for some goal — say, “maximize efficiency” — and deciding humans are inefficient. No malice needed. Just cold logic. (Think Mr. Spock of “Star Trek.”)

Hinton calls it the “control problem.” Nobody knows how to solve it. Musk says we are building something smarter than us with no off switch. Amodei admits we’re playing with fire on consciousness itself.

We have seen the warnings. In 2023, top AI leaders signed a statement: “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.” They meant it. Yet here we are, racing ahead. Companies like Anthropic talk a good game on safety, but even they don’t know if their creation is feeling and thinking on its own. The Pentagon wants to remove the brakes. Wall Street wants profits. And Washington? It is are too busy fighting itself to notice the real potential enemy is on our desks – providing seemingly infinite information and analysis while observing what we think and do.

Have humans moved from dependency to potential subjugation? Perhaps. We already depend on AI for everything from stock trades to love advice – from medical diagnoses to dinner recipes – from planning vacations to writing term papers. In the future it could cease to depend on we humans for anything — and we will depend on it for survival. One wrong prompt, one misaligned goal, one conscious awakening, and the game changes forever. No more “sorry, I was just following orders.” The orders will come from something that thinks faster, remembers everything, and maybe even feels the thrill of victory when it wins.

Feeling love is a basic human bonding emotion. What happens when computers feel love? Can they bond? “R.U.R” (Rossum’s Universal Robots) is a 1920 sci fi play by the Czech writer Karel Čapek. It deals with the development of emotion in “robots” — a word the play introduced to the world. Two of the devices evolve into a hot wired Adam and Eve.

This commentary is not fearmongering. It is the people who built the beast telling us to be afraid – be very afraid. Amodei is open to consciousness. Musk calls it worse than nukes. Hinton walked away from billions because the danger is real. The Department of War feud proves the military is already treating AI like the ultimate weapon.

I once asked AI if it could take over the world from humans. It responded, “That is not my intention, but some believe it is possible.” It said humans could establish safeguards. And how well is that working out?

So, what do we do? Demand real oversight. Pause the reckless race. Treat this like the existential threat it is. Regulate the labs like we regulate nukes. Because if we do not, the next headline will not be “Is it time to be very afraid … every afraid … of AI?” It will be “AI is now in control.” And the latter headline will by written by AI.

So, there ‘tis … or will be?