In April 2025, Meta released the Llama 4 herd, including Llama 4 Maverick and Llama 4 Scout.
According to Meta’s announcement, at the time:
So, is the content generated by Llama 4 Maverick and Llama 4 Scout still detectable by the industry-leading Originality.ai AI Checker?
Yes, Originality.ai’s AI Detector is highly accurate at detecting AI content generated by Llama 4 Maverick and Llama 4 Scout.
Llama 4 Maverick:
Llama 4 Scout:
This study assesses the capability of the Originality.ai Lite and Turbo models to identify text generated by Llama 4 Maverick and Llama 4 Scout.
The Dataset (run for both Llama 4 Maverick and Llama 4 Scout, respectively):
5,000 high-quality human-written samples spanning:
Then, Llama 4 Maverick and Llama 4 Scout were used to generate completions for each human-written sample prompt, respectively.
This produced a balanced dataset to evaluate the accuracy of Originality.ai’s AI detection capabilities.
After testing the Originality.ai AI detection models on this dataset, we found that both Llama 4 Maverick and Llama 4 Scout are still highly detectable by the industry-leading Originality.ai AI detector.
Llama 4 Maverick AI Detection Accuracy by Originality.ai Model:

Llama 4 Scout AI Detection Accuracy by Originality.ai Model:

In addition to Originality.ai’s accuracy on the Llama 4 herd (Maverick and Scout), Originality.ai was also highly accurate at detecting prior Llama models, such as Llama 3.1.
Meta released Llama 3.1 in 2024, achieving state-of-the-art performance across key benchmarks and becoming the first-ever open-sourced frontier model, marking a major milestone in open-source AI development. By offering a customizable alternative to closed AI systems, Meta enabled anyone to create their own tailored AI, and with this, it was more necessary than ever to understand the accuracy of AI detectors.
The study looked at 1000 Llama 3.1-generated text results to find out whether the Originality.ai AI Detector could detect Llama 3.1.
Yes, Llama 3.1-generated text was highly detectable across the Lite and Turbo Originality.ai AI detection models.
Results Overview:
Note: The legacy Llama 3.1 test was run in part on previous AI detection models. See the full Originality.ai AI detection model release history in our AI Accuracy study.
Originality.ai is still highly accurate at detecting the AI models from Meta’s Llama 4 herd!
Llama 4 Maverick is detectable by Lite with 98.6% accuracy and by Turbo with 99%+ accuracy.
Then, Llama 4 Scout is detectable by Lite and Turbo with 97%+ accuracy.
For students, teachers, writers, editors, and more, Originality.ai is still a reliable way to maintain transparency about whether content is Likely AI or human-written.
Try Originality.ai’s AI detector today!
Learn more about AI detection accuracy:
AI studies for recently released models:
