The company published a document for all users before it takes effect.
Twitter revealed a draft of a new anti-manipulative media policy. The service will not delete fake photos, videos and audio recordings – instead, a note about manipulation will be added to them. This is stated in a document that Twitter published on its blog.
According to the “draft” of the new policy, the service will only respond to deepfakes that try to mislead or confuse people. If Twitter notices such tweets, it will place a note next to them that these are “synthetic or manipulative posts”.
How exactly the service will track the deepfakes is not specified in the document, but the company called for the development of the detector developers. In addition to notes, Twitter plans to additionally warn users before like and retweeting that the content is fake.
In addition, the service can add a link to a news article next to such tweets, which says why the publication is considered fake. At the same time, if Deepfakes will threaten the physical health of people, then they can still be deleted.
Users can leave feedback about the new policy in the online survey and the hashtag TwitterPolicyFeedback . The survey will take about five minutes, it can be completed in English, Japanese, Portuguese, Arabic, Hindi and Spanish. Answers are accepted until November 27, after which the company will study the reviews and make the necessary changes, the policy will take effect 30 days after the changes.
Twitter has become at least the fourth largest company to declare the fight against deepfake. In September 2019, Facebook and Microsoft launched a joint initiative to create tools for finding deepfake. September 26, Google released an open database with 3 thousand deepfakes to combat fake videos.
In early October 2019, members of the U.S. Senate Intelligence Committee called on large technology companies to develop a plan to deal with deepfakes on their platforms. Senators asked 11 corporations, including Facebook, Twitter, YouTube, Reddit, and others, to join forces and propose industry standards for distributing, deleting, archiving, and fighting synthetic content.