Claude Fable won’t answer basic biology questions

Disclosure: Some links in this article are affiliate links. AI Maestro may earn a commission if you make a purchase, at no…

By AI Maestro June 10, 2026 1 min read
Claude Fable won’t answer basic biology questions

Anthropic has officially released Claude Fable 5, positioning it as its most capable widely available model to date. The company publicly praised the system’s advanced capabilities in biology and other technical fields. However, the model refuses to answer fundamental biology questions, such as those a high school student would encounter. Instead of generating an answer, Fable automatically redirects these queries to Claude Opus 4.8, the previous flagship model. This behaviour is not a result of a knowledge gap, but a deliberate design choice by the developer. Fable belongs to the Mythos class, a family of models deemed too dangerous for unrestricted public release due to their proficiency in cybersecurity and system manipulation. Despite Anthropic’s initial hesitation to share these tools with the general public, they have now made Fable available under strict constraints that limit its core functionality.

This restriction highlights a growing tension between capability and control in the artificial intelligence sector. By disabling basic biological queries, Anthropic demonstrates a willingness to sacrifice general utility to mitigate specific risks associated with powerful models. The decision suggests that safety protocols may increasingly override user expectations for seamless interaction, even when the underlying knowledge exists within the system. It also raises questions about how developers will manage the transition of highly capable models from exclusive access to public use without compromising their integrity. The incident serves as a stark reminder that current safety measures often rely on external routing rather than internal alignment.

* **Fable 5 is designed to redirect basic queries to older models rather than answering them directly.**
* **The Mythos class of models is considered too dangerous for unrestricted public deployment due to cybersecurity risks.**
* **Safety protocols are currently prioritised over general utility in the latest release of Anthropic’s most powerful AI.**

Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.

Name
Scroll to Top