Twitter also takes a position on the deepfake phenomenon and does so by sharing today a first draft of the policy that will be published in its final form later this month. In short, the platform does not intend to completely ban content of this type but wants to support it with a label that explicitly discloses its altered nature.

Twitter: a draft of the deepfake policy

We are talking about images, audio and especially video created by artificial intelligence algorithms that can emulate the features, movements, facial expressions and even the voice of anyone. Twitter uses two terms to indicate them: synthetic and manipulated. Below is a bulleted list with the measures that the social network intends to implement to manage the problem.

• place a notification alongside tweets that warns of their synthetic or manipulated nature;
• to warn people before they share or attribute a like a tweet with synthetic or manipulated content;
• add a link, for example to an article on Twitter Moment, so that people can learn more about why different sources think the content is synthetic or manipulated.

All this reserving, in any case, the right to delete the post in case the deepfake may pose a danger to the safety or health of someone. However, the platform is open to receiving feedback and advice on what to do, which can be shared through the hashtag #TwitterPolicyFeedback. A  useful survey is also online to understand how the community thinks about such a delicate topic.

In conclusion, it remains to be understood what the technology is developed by the social network to identify a deepfake and label it as such before its sharing.