Wayback Machine can be taken on a little journey through time. The students can explore their own school website, club websites or blogs. Afterwards, there should be a discussion about the advantages and disadvantages of such internet archives and what you can try as a user if unpleasant content gets onto the internet.
If private content, such as naked pictures, finds its way onto the Internet, nobody simply has to accept it. Because even if a final deletion is extremely difficult, it is still worth taking action against it! The authors can be contacted on social media platforms. If they do not remove the content, they can be reported directly if they violate the guidelines (nudity, violence, etc.). In other cases, appropriate moderators and the site operators can be contacted. With other websites and blogs, too, you should first contact the owners of the site and ask them to delete the content. If the request is not complied with, a report can be made to the police. Among other things, this can refer to the "right to one's own picture" (§ 78 UrhG) or to "pornographic depiction of minors" in the case of nude pictures of under 18-year-olds. 3
The following website is suitable for further information: www.ombudsstelle.at
Due to time constraints, this part of the module mainly deals with video deepfakes, although these only make up a small part of digital disinformation and fake news. Other exciting, in-depth aspects of manipulation in social media would be, for example, the function of social media trolls or social bots.
Deepfakes are manipulated or artificially created sound or image media that appear real. They show people who appear to be saying or doing something that they have never said or said before. Deepfakes are created using artificial intelligence such as machine learning and deep learning.
Thanks to new technological developments in the field of image processing and manipulation, deepfakes also appear more and more authentic. On the one hand, algorithms have been developed and improved in computer vision that automatically recognize and map facial structures (e.g. the position of eyebrows and nose), resulting in new technologies in face recognition. On the other hand, the triumph of the Internet - and in particular through platforms on which images and videos are shared - has created an incredibly large data pool with audiovisual data that can be used for this.
Two specific AI approaches are commonly found in deepfake programs: Generative Adversarial Networks (GANs) and Autoencoder. GANs are machine learning algorithms that can analyze a series of images and thereby create new images of comparable quality. Autoencoders, on the other hand, can extract information about facial structures from images and use that information to model a new facial expression.
Because these techniques can be used to realistically simulate facial expressions and types of movement of a person, it is now very difficult to tell whether you are looking at a deepfake or the original. However, not only the facial expressions of an existing face can be changed: Faces can be exchanged and created from scratch.
Manipulating the media and image processing are by no means new phenomena. Deepfakes are just a technological advancement of a much older phenomenon, so to speak. The emergence of social media platforms and the lively exchange and sharing of content (and thus also false content, e.g. fake news) has changed the media landscape significantly. In addition, apps like Snapchat, Instagram and TikTok already offer low-threshold filters within the applications that can be used to change faces and edit videos.
In addition, the rise of visual media, particularly video, as a means of communication is also significant. Visual media are considered to be a particularly efficient way of disseminating information. So far it was well known that misinformation is placed in texts or photos are manipulated, but video was still considered by many to be hard evidence that was difficult to forge.
False information can be spread through deepfakes, and some users can no longer distinguish between truth and fiction. Many of these reports are deliberately created to cause some form of harm. The spread of deepfakes creates uncertainty among Internet users: What is the truth and is a fact? In this case, which media can still be trusted and who is manipulating its content? Due to the sole existence of deepfakes, many users are no longer sure which content can still be trusted. 4
Deepfake technologies can be used for a variety of purposes, with both positive and negative effects. Deepfakes can be helpful, for example, in the area of audiovisual media productions (e.g. if an actor is absent), human-machine interactions can run better, but they can also find a place in areas such as video conferences, satire and art projects or surgical facial reconstruction . However, there are also a number of negative aspects, such as blackmail, defamation, bullying, identity theft, damage to reputation, manipulation of news media, loss of trust in science, business and politics, manipulation of elections, damage to international relations and national security.
Debunking deepfakes can take a long time, which means that seemingly small videos can lead to big problems. In the course of the lesson, the students should independently find examples of who uses these technologies and who they can harm. By playing through examples, you can get an idea of the dimensions that a manipulative deepfake can take on.
An apparently real video was uploaded to Instagram and Twitter of a politician confessing on camera to evading millions of euros. This video not only damages the politician's reputation and inflicts psychological damage. For example, the politician or the party could be blackmailed by threatening further forged confessions. Voters are losing trust in the party and will not vote for it again in the next election. This distrust can go so far that the system is generally no longer trusted.
Finally, in the plenum can be discussed how dangerous deepfakes can be prevented. Possible approaches would be, for example: do not post any videos of yourself on the web, avoid voice messages, do not allow unwanted ones to be recorded and insist on deleting the photos/videos, strict laws regarding. Deepfakes, stricter controls (especially on social media platforms) to curb the spread.
From a legal perspective, there are no concrete measures or laws to date. However, strategies are already being developed, such as the deepfake action plan of the Austrian federal ministries. However, no concrete changes in the legal situation are required, but awareness-raising among the population and the use of software tools that are intended to recognize deepfakes and fact-checker platforms. 5
In the last step, the students can try to create convincing deepfakes independently with the help of apps. A helpful app for this is the image manipulation app Wombo, which you can use to make selfies sing. Another possible app for this would be Reface (note: the paid Pro mode must be clicked away at the top of the X at the beginning)