Automatic spellcheckers erase typing-error biometric signal before it can be recorded. We benchmark Python spellcheckers on controlled typos and measure how much biometric utility survives.
Behavioural biometric systems based on typing errors rely on the natural tendency of people to make characteristic mistakes while typing. However, automatic spellcheckers silently correct those mistakes before the system can record them. In this work we benchmark a variety of Python spellcheckers in terms of their impact on the available biometric signal, using context-free grammar sentences with QWERTY-based and randomly generated typos, and discuss the implications for biometric system design.