Mehdi Mehdi
โ† writing
Jun 13, 2026 post #ai#anthropic

Claude Fable

Fable is a literary genre defined as a succinct fictional story that teaches a moral story. Lesson learned!

This morning, the world woke up in shock after US government decision to suspend access to Claude Fable model to non US citizens and organisations, even it’s makers (coucou Andrej Karpathy ๐Ÿง).

Anthropic decided to disable it for everyone.

Here is my best attempt at understanding what eventually led the Trump administration to shut down one of the best available model yet to humanity.

And I’ll break it down to two pieces:

First, the broken relationship between the US government and Anthropic.

In January 2026, the dispute between the pentagon and Anthropic went public after refusing to remove safeguards that blocked Claude from being used in fully autonomous weapons and mass domestic surveillance. Few weeks later, Anthropic’s CEO Dario Amodei published a statement that ultimately led to Claude being designated a supply chain risk to US national security.

The second part begun the 7th April when Anthropic launched the infamous Mythos model and Project Glasswing. Giving it to the a small group of trusted organisations and governments to patch their security holes, before a public release.

They’ve done a brilliant marketing campaign labelling it as a weapon, the thing is, when you tell the world that you’ve created a bomb long enough, they’ll end up believing you, and it backfires.

Fable was supposed to be Mythos for the general public, nevertheless, adding non deterministic safeguards to a non deterministic system is not a guaranteed solution to the security risks, and fears arise inside the pentagon. And Fable is finally disabled this morning.

I’ll be sharing my thoughts in a future post later this week.