Skip navigation
A. Flaw Found in Security System that Blocks Internet Robots

Narrator: This is Science Today. For every person who has ever taken an online poll or signed up for free web-based email, there are thousands of Internet robots, or "bots," trying to do the same thing. Jitendra Malik, a computer vision researcher at the University of California, Berkeley, explains the challenge of increasing security against "bot" programs.

Malik: This process of identifying the end of an Internet connection-is it a human or a computer-should be testable by a computer program. So it's sort of paradoxical; it's a test administered by computers, but only humans can pass it. Computers are supposed to fail the test.

Narrator: Malik recently proved that one such screening test called EZ-Gimpy, isn't secure enough.

Malik: Now it turns out that the algorithm that we developed here at Berkeley that can break EZ-Gimpy, meaning that it can solve that puzzle something like 83% of the time. So in that sense, this is not an effective test anymore.

Narrator: For Science Today, I'm Larissa Branin.