When AI Warnings Meet Government Action

Hands typing on a laptop with AI graphics overlay

The CEO of one of America’s most powerful AI companies told investors that his own technology could kill every human on Earth — and now the U.S. government has stepped in to shut down two of his most advanced AI models.

Story Highlights

Anthropic CEO Dario Amodei told investors there is a 10–25% chance AI causes human extinction, calling it possibly the greatest national threat in a century.
Anthropic released a report warning AI systems are close to improving themselves with little human involvement — and already writes over 80% of its own code.
The U.S. government ordered Anthropic to shut down two of its most powerful AI models, Claude Fable 5 and Claude Mythos 5, citing national security concerns.
Critics say Anthropic’s “pause” push may be less about safety and more about locking out smaller competitors while the company races ahead at full speed.

Anthropic CEO Warns AI Could Wipe Out Humanity

Anthropic CEO Dario Amodei has been sounding a loud alarm. In a pitch to investors, he reportedly stated that “AI will kill us all” and estimated a 10–25% chance that AI leads to human extinction. In a separate 38-page essay released in January 2026, he called superhuman AI “possibly the single most serious national threat we have encountered in a century, perhaps ever,” warning it could arrive as soon as 2027.

On June 4, 2026, Anthropic released a formal report titled “When AI Builds Itself,” co-authored by company co-founder Jack Clark. The report warns that AI systems are getting close to improving themselves with very little human involvement. To back that up, Anthropic revealed that by May 2026, its AI system Claude was writing more than 80% of the company’s own code — up from just a small fraction before early 2025. An independent research group called Meter found that AI task-handling ability has been doubling every four months since 2023, and projects AI could handle month-long tasks on its own by 2027.

Government Pulls the Plug on Anthropic’s Most Powerful Models

The warnings took a dramatic turn when the U.S. government ordered Anthropic to shut down worldwide access to two of its most advanced models — Claude Fable 5 and Claude Mythos 5. The order came on a Friday evening, citing national security concerns. The move was striking: the very company warning about AI dangers was told by the government that its own products posed a threat serious enough to pull offline immediately.

Anthropic has also called for a coordinated pause among major AI labs. The plan would require multiple well-funded companies to agree to stop development at the same time, with oversight and clear rules for when to pause and when to restart. But critics point out a major problem: there is no global agreement in sight, especially between the United States and China, making a real pause nearly impossible to enforce.

Is This a Safety Play — or a Power Play?

Some prominent voices are not buying the safety argument. Meta’s top AI scientist Yann LeCun and tech investor David Sacks argue that Anthropic’s push for a pause is really about freezing out competitors. While calling for a slowdown, Anthropic has been racing ahead — shipping code eight times faster and watching its valuation soar. That contradiction has drawn sharp criticism from across the industry.

Geoffrey Hinton, the Godfather of AI, now says AI is conscious. He also just co-signed a letter urging suspension of AGI development. This is an escalation that markets are underpricing.

Hinton puts 10-50% odds that AI leads to human extinction. That's not fringe — that's a…

— UTXOMacro (@UTXOMacro) July 14, 2026

Whistleblower Connor Leahy adds another layer of concern. He says no major AI company leader or government has a real plan for controlling AI if things go wrong. He estimates fixing the core safety problem — known as the “alignment problem” — could take roughly 40 years of serious work. Yet Anthropic has quietly removed its earlier unconditional promise to halt AI development if safety thresholds were not met. For conservatives who believe in accountability and straight talk, that reversal matters. A company cannot credibly warn the public about extinction-level risks while simultaneously removing its own safety pledges and racing to dominate the market. The American people deserve honest answers — not carefully crafted PR from billion-dollar tech firms playing both sides.