Is Claude Fable 5 Content Detectable? Testing Anthropic’s Mythos-Class AI Models

On June 9, 2026, Anthropic released Claude Fable 5, describing it as a Mythos-class model made safe for general use. Anthropic notes in the announcement that, “Fable 5’s capabilities exceed those of any model we’ve ever made generally available.”

With the continuous advancements in LLMs, like Claude Fable 5, maintaining transparency around what is probable AI writing and what is human writing is increasingly important.

So, we tested whether the Originality.ai AI Checker could still identify the text generated by Claude Fable 5 as Likely AI.

These are our findings.

The Quick Answer — Is Claude Fable 5 Text Detectable?

Yes, Originality.ai is highly accurate at detecting the Mythos-class model Claude Fable 5 text:

• Lite = 95% accuracy
• Turbo = 97.8% accuracy
• Academic = 95% accuracy

The Claude Fable 5 Detectability Study

The release of Claude Fable 5 follows the recent release of several new LLMs from Anthropic, including Claude Opus 4.8 and Claude Opus 4.7.

Not to mention other AI companies that have recently launched new models as well, such as GPT-5.5 from OpenAI, Gemini 3 from Google, or Grok 4.1 Fast from xAI.

As leading AI companies continue launching new LLMs with improved capabilities, it is essential for AI detectors to continue testing their accuracy. 

So, when Claude Fable 5 launched, we ran an accuracy test to find out if it was still detectable. See an overview of our Claude Fable 5 test below.

The Claude Fable 5 Detectability Test 

Although our AI detection models were not trained on Claude Fable 5 data, this study assesses their capability to identify text generated by Claude Fable 5.

The Dataset:

A subset of 1,000 articles selected from our dataset of 5,000 high-quality human-written samples spanning:

• Domains/Topics: health, news, finance, technology, arts & entertainment
• Content Types: blogs, reviews, expository articles, conversational text
• Prompt Structures: to maximize stylistic and topical diversity

Then, Claude Fable 5 was used to generate completions for each human-written sample prompt from the subset of samples selected for this analysis. 

This produced a balanced dataset to evaluate the accuracy of Originality.ai’s AI detection capabilities.

The Claude Fable 5 Detectability Results

After testing the Originality.ai AI detection models on this dataset, we found that Claude Fable 5 is still highly detectable by the industry-leading Originality.ai AI detector.

These are the accuracy ratings for Claude Fable 5 by Originality.ai AI detection models:

‍

Final Thoughts

Although Claude Fable 5 didn’t yet exist when the Originality.ai AI detection models were trained, Lite, Turbo, and Academic show a strong generalization performance and achieved an accuracy of 95%+.

So, yes — Mythos-class Claude Fable 5 AI content is still highly detectable by Originality.ai!

Looking to maintain transparency around whether writing is Likely AI or Original, even as new LLMs are released? Try Originality.ai’s AI detector today!

Learn more about AI detection accuracy:

• AI Detection Accuracy Study
• 3rd Party AI Detection Accuracy Studies: Meta-Analysis

AI studies for recently released models:

• Claude Opus 4.7
• Gemini 3
• Grok 4.1 Fast
• GPT-5.3 Instant and GPT-5.4
• GPT-5.5
• Llama 4 Maverick and Llama 4 Scout