This page may contain stale information. Last updated: 2026-07-01
Overview
The June 2026 US export controls on anthropic Mythos 5 and Fable 5 triggered global fragmentation — policy, commercial, and state-vendor responses to restricted frontier cyber-AI access.
Timeline
- 2026-07-01: US lifted export controls; claude-fable-5 global restore; Mythos 5 Glasswing-only (2026-07-01-anthropic-fable-mythos-5-export-ban-lifted)
- 2026-06-12: US Commerce Department suspends foreign access to Mythos 5 / Fable 5
- 2026-06-22: sakana-ai launches fugu multi-agent orchestration workaround
- 2026-06-24: 360-security unveils tulongfeng at ISC.AI Beijing
- 2026-06-27: Partial Mythos 5 restore for US critical infrastructure only
- 2026-06-28: Austria urges EU to host Anthropic domestically (alexander-proell letter)
Response Categories
| Type | Example | Feasibility |
|---|---|---|
| Policy/sovereignty | Austria EU hosting proposal | Uncertain; Commission silent |
| Commercial orchestration | Sakana Fugu | Available (excl. EU/EEA) |
| State-vendor cyber AI | 360 Tulongfeng | Claims unverified |
Key Players
- anthropic, claude-mythos
- sakana-ai, fugu
- 360-security, tulongfeng
- alexander-proell / Austria
- openai GPT-5.6 Sol (parallel US restriction pattern)
Analysis
Core tension: buying frontier AI access vs guaranteeing access under local jurisdiction. Export controls exposed dependency without fundamentally changing Europe's compute/model position.