Thematic Research

研报英文原文证据摘录

Thematic Research

Deutsche Bank

Research Institute

AI twist in the tale for

Anthropic’s Fable and Mythos

June 16, 2026

The story so far...

AuthorsIf you want to understand the implications of the abrupt switching off of

Anthropic’s Claude Fable 5 and Mythos 5 AI models globally due to US national Adrian Cox

security orders, just consider the likely reaction in the corridors of OpenAI, Research Analyst

Washington and Brussels.

The US Commerce Department issued an export control directive late on Friday

to suspend all access by foreign nationals to Anthropic’s most advanced models,

prompting Anthropic to disable access to all customers to ensure compliance.

The US said it had become aware of a method of bypassing, or “jailbreaking”,

Fable 5, the public version of Mythos 5, according to Anthropic. Mythos has only

been rolled out to a few partners because of fears that it could be misused to

exploit cybersecurity vulnerabilities. The startup disputed the concerns, which

were reportedly raised by its partner Amazon with administration officials,

saying: “We believe this is a misunderstanding.” Further details remain hazy.

Figure 1: Anthropic's Claude Fable 5 is "currently unavailable"

Source: Anthropic's Claude desktop app, accessed June 14, 2026

Anthropic is in a race for funding in what could be a historic year for technology

IPOs. After it announced on June 1 that it had confidentially filed for an IPO,

OpenAI followed suit last week. Both are expected to take place later this year at

valuations of more than $1 trillion apiece.

What could this latest twist in the tale mean for a company growing at breakneck

pace; its now smaller rival, whose ChatGPT chatbot is still the byword for AI

outside Silicon Valley; a US government that sees an existential threat from