Anthropic's Fear-Mongering Has Caught Up With Them — Eli the Computer Guy on Lost Trust in US AI

Eli Etherton (Eli the Computer Guy) joins The Tech Report to break down the latest Anthropic ban and the bigger structural question — can the world trust US AI at all? His opening framing lands hard: 'If I were an Argentinian CTO, I cannot fathom why I'd build my architecture on a US AI right now.' Worse performance is acceptable if it means 24/7 access that can't be ripped away.

This is bigger than one company. Anthropic's frontier model Mythos and the public-facing Fable are still banned, and the trigger was almost accidental — Amazon CEO Andy Jassy mentioned a jailbreak in passing during a White House meeting on an entirely different topic. From there it escalated dramatically. Jailbreaks aren't a notoriously high technical bar, and freezing top-tier models on that basis sends a chilling, unstable message worldwide.

Eli's core thesis is commercial, not technical. Anthropic's models are still strong. The problem is the marketing strategy. OpenAI built a frontier-model brand, Google built a product brand — Anthropic built a safety-and-fear brand. That same fear story is now what policymakers heard and acted on.

The 'whoever wins the AI race wins the world, because they get to own Skynet' narrative was sold by CEOs themselves. Policymakers believed it. Instead of the sober structured regulation that could have emerged, the response was impulsive, arbitrary, ban-first. Anthropic helped cultivate the very frame that's now slicing back at them.

The show also examines David Sacks's (All-In Podcast) argument: AI is good for the US economy, so as an American you should support it. Last year's AI boom added roughly 1% to GDP, Manhattan-scale data center build-outs are real demand. Coherent — but it doesn't answer the regulation question, and Eli says as much.

The global side is where the analysis bites. UK pushback against AI is small because the UK isn't building AI data centers — no factories means no opposition, but also no industry. Europe says it'll take a stronger line on the US, but whether that translates into execution is open. Either way, arbitrary US policy is hardening two trends: allies' move toward sovereign AI, and the perception that US AI products are something that can be taken away. The Anthropic / Trump friction is the first invoice on that structural cost.