Use case
ChatGPT detector for schools: detect AI writing by process
Students using ChatGPT to write essays typically paste the output rather than type it. Learnaway detects that pattern, along with uniform transcription rhythm and implausibly fast completion, without ever analysing the prose.
Catches the paste
ChatGPT output pasted into a submission window produces a distinctive large-paste event with no preceding typing. Learnaway flags and records this.
Spots transcription rhythm
Some students retype AI output. The result is a robotically uniform keystroke cadence that genuine drafting almost never produces.
Compares time to word count
A 1,000-word essay finished in four minutes is flagged with the actual submission time and word count so the teacher has the numbers.
Works on GPT-4, GPT-4o, and beyond
Because Learnaway ignores the text quality and reads only the process, it doesn't degrade as AI prose improves.
Frequently asked
- Does Learnaway detect ChatGPT specifically?
- It detects the behaviours associated with pasting AI-generated text, whether from ChatGPT, Claude, Gemini, or any other tool. Because it ignores the prose, it can't be fooled by switching models.
- What if a student types out the ChatGPT response manually?
- Uniform, low-variation typing cadence can still hint at transcription. Learnaway surfaces the cadence score alongside the other signals so the teacher can form a full picture.
- Is this a replacement for talking to students?
- No. Learnaway surfaces signals to inform a conversation, not to replace it. Its output always says 'this warrants a closer look' rather than 'this student cheated'.