As of: March 29, 2024, 7:36 a.m
By: Anika Zuschke
Comments
Press
Split
AI makes it easy to create deepfakes. This promotes a dangerous form of pornography. The victims are almost always female.
Frankfurt – If you enter “AI” on Google, the suggestions “create AI images”, “AI image generator” and “AI images for free” appear among the first search queries. The option of having a creative image tailored to your own wishes created free of charge with just one click is extremely tempting. But this world full of new possibilities also has its downsides. Because new opportunities are often accompanied by abuse. This is evident in the explosion of AI-generated deepfake pornography.
AI fuels the creation of deepfake pornography – the victims are almost always female
AI-generated deepfake pornography refers to fake pornographic photos or videos using artificial intelligence. As a study by cybersecurity company Home Security Heroes has shown, the total number of deepfake videos in 2023 was 95,820 - an increase of 550 percent compared to 2019. A full 98 percent of these videos are pornographic in nature and 99 percent Of the cases, women were victims of such deepfakes.
What is deepfake pornography?
Deepfakes are videos or photos that have been manipulated using artificial intelligence and changed to look realistic. It is a portmanteau term made up of the words “deep learning” and “fake”. Deepfake pornography is the depiction of pornographic content that has been manipulated using artificial intelligence.
The fact that these numbers have increased so much in recent years is not surprising given the hype that has arisen around artificial intelligence. AI tools are currently springing up like mushrooms and one in three deepfake tools also enables the creation of deepfake pornography. As “Home Security Heroes” describes, it now takes less than 25 minutes and is sometimes completely free to create a 60-second pornographic deepfake video of a person of whom you only have a single clear photo.
Deepfake pornography in Germany: Victims can hardly take action against it
So the problem is obvious, but what is being done about it? In the UK, non-consensual sharing – but not creation – of deepfake pornography has already been criminalized. The US has also woken up to the scandal surrounding pop star Taylor Swift, who fell victim to AI-generated deepfake pornography on X. According to
The Guardian
, a bipartisan bill was introduced in the Senate in February that would allow victims to hold accountable those involved in the production and distribution of such images. Additionally, the White House press secretary told
ABC News
about the Justice Department's first nationwide 24/7 hotline for victims of image-based sexual abuse.
A woman looks into her computer in horror - the creation of deepfake pornography has increased dramatically in recent years. © Anika Zuschke/DALL-E (AI generated)
And Germany? “It has so far been very difficult or impossible for victims of non-consensual deepfake pornography in Germany to take action against deepfake porn on the Internet,” explains Maria Pawelec, research associate at the International Center for Ethics in Science (IZEW) at the university Tübingen told
IPPEN.MEDIA
. There are regulatory gaps, among other things because it does not involve real nude photos of the victims. These would often have to rely on criminal offenses such as copyright infringement, violations of general personal rights, defamation or slander in order to take action against deepfake pornography. “This does not do justice to their suffering and the damage they have suffered,” said Pawelec.
My news
Quake in the construction industry: Read the bankruptcy of a large German real estate company
Shooting down of F-16 fighter jets: Putin threatens to attack NATO airfields read
Silent holiday before Easter – What are you allowed to do on Good Friday in Bavaria? read
Cheese recall: Federal Office warns of health risks – according to the RKI, three groups are particularly susceptible to focus reading
Nasty scam with the Rosenheim cops: Actors pack with atrocities read out in their name
She had planned Gottschalk's visit to BR completely differently: Schöneberger lost his composure
In addition, law enforcement authorities often lack the human and technical resources and expertise to even detect and take action against deepfake porn.
AI-generated deepfake images are very difficult to detect – Google and Meta have plans
In fact, a recent study at the University of Waterloo showed that the test subjects found it more difficult than expected to distinguish real people from AI-generated people. Only about 60 percent of the 260 participants could tell the difference between AI-generated people and real people - and these were completely AI-generated people, not “adapted” images of real people. “Detecting deepfakes with the naked eye is becoming increasingly difficult,” confirms Maria Pawelec. Previous errors in image generation, such as the number of fingers or a missing second earring, are becoming increasingly rare as the quality of technology increases. “Nowadays, well-made deepfakes can no longer be seen with the naked eye,” sums up Pawelec.
Google and Meta have already announced plans to label material created or modified by AI in order to make the origin of the content more clear. OpenAI – founder of the image generation model DALL-E – also wants to store both a visual watermark and hidden metadata in the generated images in order to reveal their history.
Victims of deepfake pornography “feel humiliated, powerless and intimidated”
But that doesn't change the fact that it only takes one clear picture to completely turn this person's life upside down from one day to the next. It can affect anyone – including children. “In my opinion, it has become impossible to protect yourself from becoming a victim of deepfake pornography unless you have no online presence at all,” Pawelec explains. Images from private social media accounts are often used to create AI-generated deepfake pornography.
The impact on the victims is serious: “They feel humiliated, powerless and intimidated; some even as if they had been raped. As a result, they often suffer from mental illnesses such as anxiety; There are even known cases of suicide,” says Pawelec. This is in stark contrast to the fact that, according to the study by “Home Security Heroes”, around 50 percent of the men surveyed have seen deepfake pornography at least once - and 74 percent of users do not have a guilty conscience about it.
“Politically, more definitely needs to be done against deepfake pornography,” says Pawelec. But it also holds accountable the platforms that benefit financially from the distribution of non-consensual deepfake pornography, as well as companies that enable users to create deepfake porn.
EU agrees on new directive that criminalizes the distribution of deepfake porn
In fact, the European Commission provided a ray of hope in March 2022 when it proposed a directive to combat violence against women and domestic violence. The European Parliament and the Council agreed on the directive in February this year, which also criminalizes online violence - including the non-consensual exchange of intimate deepfakes. “I very much welcome this and hope that the directive will soon be implemented at national level,” said Pawelec.
Ursula von der Leyen, President of the European Commission, commented on the directive on is an important step against many forms of violence in the real world, but above all it brings with it profound changes for the online world." It is a step in the right direction, you are there with a view to the extent of AI-generated Deepfake pornography still has a lot to do on the part of the platforms and companies that allow the creation and distribution of such deepfakes.