Claude Fable 5 access restored on AI Gateway
Executive Take
A US export control briefly cut off access to a frontier AI model overnight, then restored it days later. Any company building on a single model provider now has a real case study in geopolitical dependency risk.
Executive Summary
Vercel restored access to Anthropic's Claude Fable 5 model on its AI Gateway after the US government lifted export controls that had blocked it. The model is unchanged from its June 9-12 release, but safety classifiers are now more robust. Fallback models can be configured if Fable 5 refuses a request. Anthropic does not offer Zero Data Retention for this model, keeping prompts for 30 days.
Why It Matters
Technology and AI leaders relying on Claude models for production workloads need to know export controls can suddenly restrict access. This is also a reminder to check data retention policies before routing sensitive prompts.
Bizquad Perspective
Most teams treat model access as a stable utility, but this shows AI infrastructure can be switched off by government policy with no warning, so fallback plans belong in every AI deployment, not just an engineering nice-to-have.