Most AI writing detectors work by sending your text to a remote server for analysis. Vyasa takes the opposite approach: it runs entirely in your browser using WebAssembly (WASM) - a technology that lets code run at near-native speed inside a web page - and never makes a single API call. Your text stays on your machine.
The tool is built around Wikipedia's "Signs of AI Writing" guide, a checklist that Wikipedia editors developed after the platform banned undisclosed AI-generated articles. Vyasa takes those human-identified patterns and turns them into an automated scanner that checks for five categories of red flags: AI-typical vocabulary, structural patterns, promotional or legacy phrasing, self-referential language, and other stylistic tells.
It handles text up to 1 million characters, which covers most documents anyone would need to check. Results show a breakdown of flagged phrases with explanations of what each pattern indicates and why it's associated with AI output.
What It Gets Right - and What It Doesn't Claim
Vyasa is refreshingly honest about its limits. The tool explicitly warns that its findings are "indicators, not proof" and that every flagged pattern can also appear in human writing. That's the correct framing - something many commercial AI detectors with 99% accuracy claims fail to mention.
The detection approach is pattern-matching, not statistical classification. It's looking for the crutch words and structural habits that large language models lean on - the kind of tells that experienced editors learn to spot. Think of it as an automated version of a trained human reviewer scanning for red flags, not a black-box classifier making probabilistic guesses.
The privacy angle is the real selling point. Teachers checking student work, editors reviewing submissions, or anyone handling sensitive text can run checks without that content hitting a third-party server. For organizations with strict data handling requirements, that's a meaningful difference from tools like GPTZero or Originality.ai that process everything server-side.
Vyasa is free and open source. It won't replace a careful human review, and it shouldn't be used as the sole basis for accusing anyone of using AI. But as a quick first-pass screening tool with zero privacy cost, it fills a gap that the commercial detectors have ignored.