Fable 5 Was So Good, The US Goverment Banned It Within 76 Hours
Last Updated on June 18, 2026 by Editorial Team
Author(s): Caspar Bannink
Originally published on Towards AI.
Fable 5 Was So Good, The US Goverment Banned It Within 76 Hours
Anthropic's Claude Fable 5 launched on June 9, 2026, and was so capable that the US government issued an export control directive to suspend foreign-national access 76 hours later. Anthropic publicly disputes the directive’s basis, citing a “jailbreak” capability that is “widely available from other models (including OpenAI’s GPT-5.5).” The model landed #1 on the Artificial Analysis Intelligence Index at 64.9, ahead of GPT-5.5 xhigh at 60.2, Claude Opus 4.8 max at 61.4, and Gemini 3.1 Pro Preview at 57.2. The reason the launch produced 76 hours of public access, four playable browser games, an interactive cartographic webapp, a 3D learning experience, and a working data-analysis UI, all from single-prompt Claude Code sessions, is the same reason it got banned.

After the lead, the article argues that Fable 5’s “banned so fast” story is best explained by concrete, real-world outputs produced in days: multiple sources independently describe one Claude Code session generating several finished browser games (including Flipside, Snake, Strata) plus a working cartographic webapp and data-analysis UI, with detailed game mechanics and UX elements rather than demos. It adds more evidence from other builders—Koen van Gilst’s Shepherd’s Dog built in about 45 minutes, Shinya Hara’s 3D learning island app, and a community HTML5 shooter released on launch day—then contrasts these demos with the benchmark picture, noting that the main independent performance ranking is from Artificial Analysis (where Fable 5 leads) while public frontend/UI evaluations still haven’t caught up. The author concludes by explaining what to test in your own stack (tool-call counts, time-to-finish, one-prompt app builds, and rerouting/classification frequency), and by detailing what the 76-hour export-control window effectively enabled for builders: a wave of shipping, playable artifacts that outran both the regulatory response and the broader benchmark coverage.
Read the full blog for free on Medium.
Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI
Towards AI Academy
We Build Enterprise-Grade AI. We'll Teach You to Master It Too.
15 engineers. 100,000+ students. Towards AI Academy teaches what actually survives production.
Start free — no commitment:
→ 6-Day Agentic AI Engineering Email Guide — one practical lesson per day
→ Agents Architecture Cheatsheet — 3 years of architecture decisions in 6 pages
Our courses:
→ AI Engineering Certification — 90+ lessons from project selection to deployed product. The most comprehensive practical LLM course out there.
→ Agent Engineering Course — Hands on with production agent architectures, memory, routing, and eval frameworks — built from real enterprise engagements.
→ AI for Work — Understand, evaluate, and apply AI for complex work tasks.
Note: Article content contains the views of the contributing authors and not Towards AI.