Stylome · ChatGPT integration

Evidence about a passage’s writing process.

Ask Stylome to analyze an English prose passage for patterns consistent with human writing, AI editing, or AI generation. Results include relative confidence and clear limitations.

The integration supports five successful scans per OpenAI-provided user identifier per UTC day. Passages must contain 200–10,000 characters after whitespace normalization. Errors and benchmark lookups do not consume the scan allowance.

Understand the evidence

Ask for the published benchmarks, including false positives, detection recall, source and producer breakdowns, and study limitations. Read the benchmark feed or the full transparency reports.

Classifications are statistical evidence, not proof of authorship. Confidence is not a probability of correctness. Do not use a result alone for discipline, hiring, or other consequential decisions.

Data handling

We process the passage you choose. We do not store the text, its hash, or its scan result in this integration. We keep short-lived, one-way daily quota counters. No Stylome sign-in is required.