Twitter said last month it was introducing a new policy to help crack down on fake photos, as well as “manipulative media” including photos, video and audio, which apparently changed its original meaning and purpose, or appeared to have happened something. Today, Twitter unveiled a draft of the new policy, hoping to consult the public before it is officially released.
The new policy aims to address the growing Problem of Deepfake on the Internet. Deepfake is, in layman’s terms, AI face-changing, a branch of artificial intelligence. The 2017 Deepfake video caused a stir when deep learning began to introduce A’s face to B’s face, a behavior that caused public outrage and the government began studying legislation banning it.
With artificial intelligence, Deepfake can also extend to many media, such as video, audio, and photos, so what you see isn’t necessarily real. To that end, in July, Virginia introduced an amendment that would make it illegal, whether real or false, to share nude photos or videos of a person’s video without permission, including images of Deepfake. The violation of the law is a category 1 misdemeanor punishable by up to 12 months in prison and a fine of up to $2,500.
In early October, two members of the Senate Intelligence Committee, Mark Warner (D-VA) and Marco Rubio (R-FL), called on big technology companies to develop a plan to crack down on fake content on their platforms. Senators have asked 11 technology companies, including Facebook, Twitter, YouTube, Reddit and LinkedIn, to come up with a plan to set industry standards to “share, delete, archive and respond to synthetic content sharing as soon as possible.”
Twitter announced its plans later this month to seek public feedback on the policy. Meanwhile, Amazon has partnered with Facebook and Microsoft to support the DeepFake Detection Challenge (DFDC), which aims to develop new ways to detect manipulative media.
Today, Twitter is detailing a draft of its Deepfake policy. The company says it will:
Place notifications next to tweets that share compositions or manipulate media;
Warning people before they share or like tweets through synthetic or manipulative media
Add a link (for example, to a news report or Twitter Moment) so that people can read more about why various sources think the media is synthetic or manipulating the media.