Twitter drafts a deepfake policy that would label and warn

Twitter drafts,deepfake policy that would label

Today, Twitter is detailing a draft of its deepfakes policy. The company says that when it sees synthetic or manipulated media that’s intentionally trying to mislead or confuse people it will:

  • place a notice next to Tweets that share synthetic or manipulated media;
  • warn people before they share or like Tweets with synthetic or manipulated media; or
  • add a link – for example, to a news article or Twitter Moment – so that people can read more about why various sources believe the media is synthetic or manipulated.

Twitter says if a deepfake could threaten someone’s physical safety or lead to serious harm, it may also remove it.

The company is accepting feedback by way of a survey as well as on Twitter itself, by way of the #TwitterPolicyFeedback hashtag.

The survey asks questions like whether altered photos and videos should be removed entirely, have warning labels, or not be removed at all. And it asks whether certain actions are acceptable,

like hiding tweets or alerting people if they’re about to share a deepfake. It also asks when it should remove a tweet with misleading media. The policy Twitter created says tweets will be removed if the tweet threatens someone’s physical safety,

but will otherwise be labeled. The survey suggests some other times a tweet could be pulled — like if it threatens someone’s mental health, privacy, dignity, property and more.

The survey takes five minutes to complete and is available in English, Japanese, Portuguese, Arabic, Hindi and Spanish.

Twitter drafts,What isn’t clear, however, is how Twitter will be able to detect the deepfakes published on its platform

given that detection techniques aren’t perfect and often lag behind the newer and more advanced creation methods. On this front, Twitter invites those who want to partner with it on detection solutions to fill out a form.

Twitter is accepting feedback on its deepfakes policy from now until Wednesday, November 27 at 11:59 p.m. GMT. At that time, it will review the feedback received and make adjustments to the policy, as needed. The policy will then be incorporated into Twitter’s Rules with a 30-day notice before the change goes live.