Proposal of CAPTCHA resistant to relay attack
Tokuichiro Suzuki, Takumi Yamamoto, Masakatsu Nishigaki · 2010
It has been recently reported that malicious users who attack CAPTCHAs are gradually changing their strategy from using automated programs to using human solvers. Such malicious users try to bring net-surfers around the world together by hosting some attractive web site. The malicious web site accesses a victim web site to obtain a CAPTCHA test on the victim site. Then, the malicious site relays the CAPTCHA test to net-surfers who are visiting the malicious web site. The CAPTCHA test will be solved by the net-surfers, and thus the malicious web site can send the CAPTCHA response to the victim site. This is how the malicious web site can use those net-surfers as human resources to solve CPTCHA test on victim web sites. These kinds of attacks are called relay attacks. So far, many researches have studied to improve CAPTCHAs’ tolerability against a various attacks conducted by automated programs (malwares). Those countermeasures will, however, not work at all, since the attackers are human beings in the relay attacks. Therefore, we urgently have to tackle the relay attacks. This paper focuses on the difference in PC between the entity who accesses the victim site and the entity who solves the CAPTCHA test under the circumstance relay attacks are conducted. Based on the observation, we propose a relay attack detecting scheme by comparing the IP addresses of the accessing entity with that of the solving entity. 1. はじめに