This is welcome news for Twitter users who feel that Twitter’s visual recognition algorithms have penalized them for not detecting unfairly detected content.
Today Twitter is launch a new test an updated appeals process for breaches of sensitive media to make it easier to question such decisions.
Starting today, we’re testing a way to complain if you think we’ve mistakenly added your tweet to include sensitive media.
Your feedback will help us build a fairer and more transparent process. (1/2)
– Twitter Support (@TwitterSupport) August 3, 2021
The process provides a more direct way to complain about the operation of tweets and helps users get a response from Twitter if their content is incorrectly flagged.
Twitter’s image recognition processes have been the cause of several high-profile events classification error on this front. Just last month, the stage came on fire as a picture of a badly returned army veteran greets the uniform on July 4th. marked for the inclusion of “potentially sensitive content”.
Realistically, Twitter must rely on identifying artificial intelligence to limit the distribution of harmful content as much as possible, and its systems, Twitter says, will lead to improvements in this regard.
In the latest Transparency report, Twitter reported Impression-breaking tweets – those that violate Twitter rules for any reason – accounted for less than 0.1% of all tweet views during the period, and its artificial intelligence detection tools help significantly limit exposure to such over time. Twitter also said Its number of violations of sensitive media practices increased by 322% and the number of non-consensual nudity accounts increased by 194%.
Overall, these are good results, but increased activity and reliance on automatic preventive restraint also lead to more false positive results.
Therefore, this improved complaint process makes sense, and hopefully will allow Twitter to continue optimizing its detection systems while serving users who have been wrongly penalized.
The first test of the new complaint process will be available to a small group of users in Mexico, the United States and Japan, the latter two of which are Twitter’s largest user markets.
“If you’re in an experiment and haven’t marked your own media as sensitive in the settings, you can complain about a tweet marked as tweet itself (network only) or a tag notification.”
It’s a good update that responds to Twitter’s broader shift toward increasing user protection.