We're frozen out (for good?)
TL;DR
Frontier AI is now restricted: OpenAI was asked to hold back GPT 5.6, and Anthropic pulled Mythos and Fable after government pressure, fulfilling predictions from Leopold Aschenbrenner's situational awareness paper about 2026-2027 restrictions.
A two-tier AI system is emerging: Shapiro predicts a permanent split between frontier models reserved for government, military, and top-paying corporate clients, versus general-purpose models for the public.
This is Cold War 2.0: The US-China rivalry now centers on AI, humanoid robots, and upstream technologies like chips and power, shaping all investment decisions.
Cybersecurity is the key argument for access: Shapiro argues that powerful AI models would actually strengthen cybersecurity by automating penetration testing, auditing, and best practices, helping defenders more than attackers.
The GPS analogy offers hope: GPS was initially deliberately degraded for civilians, but the government eventually removed restrictions when the strategic advantage faded, suggesting AI access could follow a similar path.
The Breakdown
The US government has asked OpenAI to hold back GPT 5.6 and pressured Anthropic to pull its frontier models Mythos and Fable, signaling what Shapiro sees as a permanent bifurcation between weapons-grade AI for government and top-tier corporate use, and watered-down models for everyone else.
Was This Useful?
Share
Keep Reading
Make Alcreon Yours
Tune your feedFive quick questions, and the feed ranks what matters to you first.Or just get notified
The weekly Echo. Signal worth keeping in your inbox.
Every new piece, announced on X.
Read Next
See all
Playbook
The Cheapest Model That Passes
OpenRouter lists 400 models behind one API. The fix for choosing isn't a better leaderboard, it's a four-step protocol that ends in a real eval.
Playbook
Cheap Models, Hard Tasks
Most agent workflows route every step to the frontier model by default. The bill scales with how chatty the agent gets, even when most steps don't need that brain.
Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.