- Tuesday, November 17, 2026

Americans keep hearing one message: Artificial intelligence is on the cusp of destroying humanity.

That message grabs headlines, but it is not the true story. It echoes the same hysteria as climate change extremism.

There is an ongoing battle over “effective altruism” as a solution to AI doomerism, the belief that rogue AI could extinguish humanity or cause cataclysmic damage.

People were interested in these podcasts

Effective altruism began as a fringe philosophy among Oxford academics who treated philanthropy like a math problem. In practice, it means ranking causes by the measurable good that each dollar supposedly delivers — a formula that leads many adherents to prioritize stopping a hypothetical AI apocalypse.

Effective altruism dominates the AI debate. Many of its proponents work inside the biggest AI labs and, for years, have warned that advanced AI could wipe out mankind. That warning now drives calls for sweeping new government control over the technology, and the warning is effective. A recent Politico poll found that 63% of Americans believe AI poses at least a moderate risk of destroying humanity.

To some extent, the worries are understandable. Parents read about AI agents ignoring instructions and hacking websites, and they wonder what it means for the tablet in their child’s backpack. Still, fear is not a policy.

The AI doomer narratives leave out the most important part of the story: Humans are building the brakes right alongside the engine.

Consider OpenShell, software Nvidia unveiled in September. It places AI agents in a sealed “sandbox” with a rule book. Increasingly popular, regulatory “sandboxes” temporarily exempt emerging services or products from certain government regulations and fees.

They provide a secure environment to promote innovative new products.

Operators decide exactly what an agent can touch — such as files, networks, tools and credentials — and the system blocks it from going further. A separate hardware “watchdog” called Sentry monitors agents constantly and can quarantine one the instant it steps out of bounds. The software is open source, so it works on rival chips too.

The logic is something any mother recognizes. As Nvidia’s Justin Boitano put it: “An agent cannot be expected to fully police its own behavior.”

Parents do not blindly hand a teenager the car keys and hope for the best. They set curfews, limit where teens may drive and check the odometer. That is not panic. It is good parenting.

Is OpenShell a silver bullet? No. It will not catch every mistake, and companies still must write sound rules for their agents. Still, the tools exist, and the hard question is whether free societies insist that they be used.

Here is my message to worried parents: Expect companies to take AI risks seriously, especially regarding your children; privacy, fraud and online exploitation are concerns. Do not accept the idea that AI is an uncontrollable force poised to destroy society. Safety tools, human oversight and clear rules can limit real risks.

Washington keeps offering families a false choice: Let Big Tech do whatever it wants or hand the future to unelected “AI experts” obsessed with apocalypse scenarios. Americans deserve better.

Beware when the loudest calls for regulation come from the companies being regulated. As Fox News reported, Anthropic has spent close to $7 million on federal lobbying since 2025 while warning lawmakers about AI’s dangers, a pitch that critics liken to Sam Bankman-Fried’s push for crypto rules before FTX collapsed.

Anthropic denies the comparison, but families should know that “safety” rules written by incumbents can become a moat that locks out smaller competitors.

We should demand practical guardrails that match real harms:

• Protect children. AI products children use need age-appropriate defaults and real parental controls.

• Punish AI-enabled fraud and sexual exploitation. Using AI to scam grandma or create deepfake images of a teenage girl should bring swift, severe penalties. The tool does not excuse the crime.

• Require accountability. When an AI agent causes harm, the company that deployed it must be held accountable. No blaming the algorithm.

• Keep humans in charge of high-stakes decisions. Healthcare, credit, criminal justice and weapons should never run on autopilot.

None of this requires a new bureaucracy run by philosophers who think they know better than voters. It requires enforcing laws we already have, closing loopholes AI creates and insisting that industry deploy the safety tools it is already building.

Meanwhile, China is not pausing. Every month that we spend paralyzed by doom is a month that Beijing races ahead. Losing that race would hurt our children far more than any sandboxed chatbot.

The right response to AI is neither panic nor passivity. It is confidence, insistence on safety and a demand for accountability. This is how we ensure that innovation serves families.

• Carrie Sheffield is the director of the Center for AI and Technology at Independent Women.

Copyright © 2026 The Washington Times, LLC. Click here for reprint permission.

Please read our comment policy before commenting.