Twitter introduced a new bounty program that will reward DEF CON AI Village participants who discover and disclose signs of bias in the company’s image cropping algorithm before August 6.
The algorithm in question is used to determine which part of an image should be shown in Twitter’s mobile apps, and the company said in May that it had received complaints that the algorithm “didn’t serve all people equitably” because of “gender and race-based biases.”
Twitter’s analysis backed up those complaints: It said “an experiment of randomly linked images of individuals of different races and genders” that was conducted on 10,000 images showed that its image cropping algorithm was indeed biased in favor of women and white individuals.
The company responded by giving Twitter users direct control over how an image would be cropped in mobile timelines before it was sent, which it said “reduces our dependency on [machine learning] for a function that we agree is best performed by people using our products.”
Now Twitter is challenging DEF CON AI Village participants to find similar problems in the same algorithm. The company said that “with this challenge we aim to set a precedent at Twitter, and in the industry, for proactive and collective identification of algorithmic harms.”
Participants were told to examine Twitter’s image cropping algorithm and the code on which it relies in an effort to “demonstrate what potential harms such an algorithm may introduce” and could affect “anyone from Twitter users to customers or Twitter itself” in the process.
The program will offer between $500 and $3,500 in cash prizes to five winning teams who will present their work on August 6. More information about the program is available via HackerOne.
