The Rip Current with Jacob Ward

The Rip Current with Jacob Ward

Seven Things AI Can Already Do That Should Terrify You

The Pentagon just tried to strong-arm Anthropic into removing its AI safety guardrails. Here’s a few examples of the kind of capabilities those guardrails were holding back.

Jacob Ward's avatar
Jacob Ward
Feb 26, 2026
∙ Paid

Editorial Note: I’ve been trying to stay away from scaring my audience — I did a whole book in that voice, and I know it’s not for everyone. But as Aza Raskin of the Center for Humane Technology put it when he was on the show in July: there’s plenty of work being done right now that will fuck up your sleep. As the national conversation increasingly normalizes AI’s role in our lives, and as we approach Anthropic’s Friday deadline from the Pentagon, it’s worth laying out what the technology can actually do right now, and why you should worry about it.


When Defense Secretary Pete Hegseth summoned Anthropic CEO Dario Amodei to the Pentagon and demanded an unguardrailed version of his company’s AI — reportedly threatening to declare Anthropic a “supply chain risk” if it refused — the press covered it as a policy dispute. A philosophical difference about AI safety norms versus national security imperatives.

The Trap Closes on Anthropic

The Trap Closes on Anthropic

Jacob Ward
·
Feb 24
Read full story

But that framing skips an important question: What exactly are the capabilities the Pentagon wants? What can AI do that Anthropic is worried about?

The capabilities that AI already has — documented in peer-reviewed research, military procurement records, and live deployments around the world — go far beyond what most people picture when they hear the term “artificial intelligence.” We are not talking about chatbots writing convincing propaganda or ripping off musicians. We are talking about systems that see through walls, identify you by your heartbeat, and run autonomous kill chains from detection to detonation with no human pulling the trigger.

Security cameras monitor the surroundings from all angles.
Photo by Carl Tronders on Unsplash

What follows is not speculation. Every item below is sourced from published research, military records, or credible reporting. This is a field guide to what AI can already do — and a hint at what a defense secretary might want to do with it if he could strip out the safety rails.

Paid subscribers: read on for the full rundown — seven documented capabilities that most people have no idea exist yet.

User's avatar

Continue reading this post for free, courtesy of Jacob Ward.

Or purchase a paid subscription.
© 2026 Jacob Ward · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture