Frequently asked questions

Honor Checker provides deep, language-aware analysis for 14 languages: Java, Python (2 and 3), JavaScript, TypeScript, C, C++, C#, Go, Rust, Kotlin, Swift, Dart, and PHP — plus a generic fallback that supports any text-based source format. Language-aware parsing (ANTLR-based, with variable normalization) understands code structure rather than treating it as plain text.

Yes. Comparisons span every section and academic period of a course, so a submission recycled from an earlier term — or borrowed from another section — still surfaces.

Instructor-provided starter code is fingerprinted and excluded, and a “de facto boilerplate” threshold automatically suppresses code that appears across many submissions. That means shared scaffolding doesn’t turn into false positives.

No. Before fingerprinting, code is normalized via AST transformation, so renamed variables, reformatted whitespace and reordered imports don’t defeat detection. Honor Checker compares structure, not surface text.

No. Detection is based on fingerprints of the code, and a fingerprint matches regardless of where it appears in a file. Moving functions around, shuffling the order of methods, or rearranging blocks doesn’t hide the shared logic — the same fingerprints still match wherever they end up.

Honor Checker runs on AWS with a serverless architecture and per-tenant data isolation — every record is scoped to its institution. Authentication is handled by AWS Cognito with required multi-factor authentication, and data is encrypted in transit. See our Security & privacy page for details.

Access is invitation-based and role-scoped. Organization, course and section admins invite co-instructors and staff with the appropriate scope, matching how your institution and courses are organized.

Yes — Honor Checker integrates with Instructure’s Canvas, so instructors can bring submissions in directly rather than uploading them by hand. Additional LMS integrations will be added based on demand. You can also always upload single submissions or submission batches by course section. If your institution uses another LMS, let us know via the contact form so we can prioritize it.

Honor Checker uses institutional licensing, priced by institution or department size. There are no per-seat fees, and detailed pricing conversations begin at general availability. See the Pricing page for the model.

Detection today is a deterministic, published algorithm — not AI. AI assistance features are on the roadmap, and every one of them follows a single principle: deterministic evidence first, AI assistance second, and a human always decides. AI never accuses a student.

No, and we won’t ship verdict-style AI-authorship detection. The current science of AI-text detection has a poor false-positive record, and accusing a student on that basis isn’t defensible. What does help: students who share the same AI-generated solution with each other are still caught by cross-submission matching, and our roadmap’s comprehension-interview questions help instructors verify authorship the defensible way — by talking to the student.

Beta means the core detection is complete and doing real work — it’s in production use grading actual courses at Brigham Young University — while we finish additional features, including AI assistance, ahead of general availability. The pilot is not open to additional institutions yet. Join the notification list and we’ll email you at general availability.