Originality
Research Lab.
We are an independent forensic linguistics center developing the integrity layer for the next generation of academic software.
Bridging the Gap Between Probability and Plagiarism.
Traditional academic integrity tools were built for a “copy-paste” world. The generative AI era requires a fundamentally different approach: analyzing the statistical entropy of language itself.
Our mission is not to police students, but to empower institutions with transparent, verifiable forensic data that distinguishes between human creativity and algorithmic output.
Every score our system produces comes with a full confidence breakdown — not just a verdict, but the reasoning behind it. We believe academic integrity decisions should be supported by evidence, not black-box outputs.
model
Evolution of Our Detection Models
How our linguistic engine has developed against the rapid advancement of generative AI.
Initial development of entropy analysis in response to the widespread availability of large language model writing tools. Early focus on basic perplexity scoring and establishing baseline human writing signatures across academic disciplines.
Integration of sentence variation metrics — burstiness analysis — to reduce false positives in ESL student papers and formal academic writing. This model significantly improved discrimination between non-native human writers and AI-generated text with uniform syntactic patterns.
Full LTI integration with major learning management systems. Introduction of paragraph-level confidence scoring and vocabulary distribution analysis, enabling instructors to identify exactly which sections of a submission triggered the AI signal rather than relying on document-level averages.
Extended detection coverage to 14 languages. Separate perplexity baselines were established for each language to avoid penalizing international students whose native-language writing patterns differ from English academic norms. Humanizer evasion detection added as a dedicated layer.
Real-time model retraining against the latest generation of AI writing tools and humanization techniques. Our detection dataset is updated continuously as new AI model releases change output characteristics. Active collaboration with institutional partners to collect edge cases and improve accuracy across specialized academic domains.
Advisory Council
Guided by leaders in Computational Linguistics, AI Ethics, and Academic Integrity.
“Our goal is transparency. We interpret the statistical noise so educators don’t have to make unsupported judgment calls.”
“We advocate for a human-in-the-loop approach. AI detection is a tool that informs decisions — it should never be treated as a judge.”
“Detecting AI is an ongoing challenge. Our models retrain continuously to match the pace at which language models themselves evolve.”
Join the Research Consortium
Partner with us to contribute to the next generation of academic integrity research. Consortium members receive early access to new detection models and priority support.