How do answer-scraping bots harvest exam content?
Short answer: Scraping fleets enroll in or access test-prep and assessment platforms, then page through question banks at machine speed, screenshotting or extracting every item. The harvested bank is sold as 'study material' or used to pre-answer live exams. Once the bank leaks, every future sitting of the exam is compromised.
The leak path nobody watches
Platforms protect the live exam window and forget the practice interface, the review mode, the API behind the results page. Bots harvest where the security camera is not pointed - often the same content, minus the proctoring.
Why watermarking alone fails
Watermarks identify the leaker after the leak. They do nothing to stop the harvest, and farms rotate disposable accounts precisely so attribution leads nowhere. Prevention has to happen at read time, not after resale.
Rate is the tell
A human reviews a question in tens of seconds and rereads some. A harvester moves at fixed machine intervals with perfect coverage and zero hesitation. Scoring read behavior across the bank session catches the harvest while there is still something left to protect.