DualCheck: Exploiting Human Verification Tasks for Opportunistic Online Safety...
summary
DualCheck embeds 10 multiple-choice online-safety questions into a CAPTCHA-like flow and collects cursor behavior to run verification similarly to reCAPTCHA v2 (actual reCAPTCHA code is not integrated). A 15-day deployment with 34 participants shows mean accuracy on the question set rises significantly from day 1 to day 15, no significant difference between one-time and repeat-until-correct modes, and higher perceived usability (SUS) than text- or image-based CAPTCHAs.
AI-written summary · about
chapters (11)
- 0:00Intro
- 0:26Background Issues on online safety literacy
- 2:21System design
- 3:02Prototype
- 4:02Question Curation Prepared 10 questions
- 4:512 modes of DualCheck
- 5:27Procedure
- 6:16Mean accuracy rose after the study
- 7:03Positive learning effect of original questions
- 8:11Limitations and future work Limitations and future work
- 9:27Summary Thank you for listening!
related talks
16mUser (Non-)Compliance with Age Verification: Evidence from a Deceptive Web Experiment
28mI'm Not a Human: Breaking the Google Recaptcha
05mSenCAPTCHA: A Mobile-First CAPTCHA Using Orientation Sensors
05mWebAlly: Making Visual Task-based CAPTCHAs Transferable for People with Visual
01mCAPTCHA defeat
16mChecking, nudging or scoring? Evaluating e-mail user security tools
01h 15mStrengthening trust in computing technologies for detecting illegal deception online
40mAdaptive Testing Methodology: Crowdsourced Testing...
