Translation: The digital doppelganger dilemma:
AI-generated intimate content and its malicious implications
A critical look at deepfakes, their alarming effects and how we can protect our reality.
This article describes the rise of deepfakes and AI-generated intimate content and highlights their development, current impact, legal responses, future development and possible individual protective measures.
The problem: deepfakes and AI-generated intimate content
Definition Deepfakes are AI-generated explicit scenes or manipulated images that show people, often without their consent. This technology goes beyond simple face swapping and enables the creation of completely fabricated explicit content.
Nature of the threat The content is hyper-realistic, non-consensual and predominantly targets women. It represents a „distorted mirror“ in which the image of individuals is twisted and used against them.
effects Profound, gender-specific digital violence: Leads to severe emotional trauma, anxiety, reputational damage, cyberbullying, blackmail and potential legal consequences in the real world for victims.
A Brief History of Forgeries
Early phases (1990s–2000s) Researchers experimented with facial reanimation (e.g. the Video Rewrite Programme in 1997) and developed fundamental AI technologies such as autoencoders and early neural networks.
The Game Changer (2014) The development of Generative Adversarial Networks (GANs) marked a turning point that turned the creation of realistic fakes from a theoretical possibility into reality.
Public disclosure and democratisation (from 2017) The term „deepfake“ was coined on Reddit and quickly became associated with pornography in which the faces of celebrities were swapped (Celebrity Face Swap). Easy-to-use tools such as FakeApp democratised the creation of convincing fakes and enabled individuals with malicious intentions to create them.
Exponential growth Since 2019, deepfake content has doubled every six months, indicating an exponential increase in digital deception.
The Alarming Present: Current Landscape
Prevalence of non-consensual pornography 96–99 % Online deepfake videos are non-consensual pornography, whereby 99 % target women.
Limitations of human perception Humans only have an accuracy of 55–60 % in detecting deepfakes.
Public apathy A significant portion of the public, particularly young men, either sees no harm in the creation/dissemination of this content or is neutral towards it, suggesting a collective failure of empathy and critical thinking.
Real-world incidents
Taylor Swift deepfake (early 2024): Demonstrated the vulnerability of even global icons to widespread, harmful content.
Bullying in the school playground: Deepfakes are used as a weapon for peer harassment, primarily targeting girls.
Political disinformation: Fake Zelensky videos and misleading political advertisements spread disinformation.
Health misinformation: AI-generated doctors spread false health advice.
Speculation about „19-minute viral video“ (December 2025): Demonstrated how quickly unconfirmed content can lead to public harassment.
Hong Kong incident (2025): University students discovered AI-generated pornography of themselves, highlighting the personal, real-world implications.
The Law Catches Up: Global Legislative Responses
Germany
Extends existing laws (data protection, image rights, defamation), which are often insufficient.
Proposed criminal offence (Section 201b of the Criminal Code, draft July 2024) for „violation of personal rights through digital forgeries“.
In June 2024, a bill was introduced to criminalise deepfakes without consent.
German Parliament: Draft law on deepfakes
European Union
Digital Services Act (DSA, November 2022): Increases the accountability of platforms for illegal content, including deepfakes.
EU AI Act (August 2024): Classifies deepfakes as „low risk“ but mandates transparency (mandatory labelling of AI-generated content). Critics consider this insufficient for explicit deepfakes.
Directive on combating violence against women (May 2024): Requires all EU Member States to criminalise non-consensual sexualised deepfakes by June 2027.
Data Protection Foundation: EU Directive on Digital Violence
United States
TAKE IT DOWN Act (May 2025): The first federal law that criminalises the publication of non-consensual intimate deepfakes, with a 48-hour „notice and takedown“ procedure („notice and takedown“) for platforms until May 2026.
State laws: 27 Some states have laws against sexual deepfakes; others deal with election-related content, with varying definitions.
Future federal proposals: Deepfakes Accountability Act (digital watermarks), NO FAKES Act (unauthorised voice/image), Protect Elections from Deceptive AI Act, Deepfake Liability Act (reform of Section 230), Federal Rule of Evidence 707 (AI evidence standards).
Other countries Denmark (image as intellectual property), France (criminal liability, labelling), China (consent, labelling), United Kingdom (Online Safety Act).
Challenges in enforcement Anonymous authors, cross-border content and the sheer volume of material make enforcement difficult.
The Never-Ending Arms Race: Outlook for the Future
Advances in deepfake technology
Higher resolution, flawless movements and fewer noticeable „features“.
Live deepfake generation, creation of entire synthetic scenarios, and interactive digital humans.
Increased accessibility of sophisticated counterfeiting tools on smartphones.
Advances in recognition technology
AI and ML for more accurate, autonomous detection systems.
Explainable AI (XAI) to clarify detection methods.
Multimodal and real-time recognition across video, audio and physiological cues.
Blockchain and watermarks for digital fingerprinting and content authentication.
Focus on voice deepfakes, voice cloning and sophisticated vishing attacks.
Evolving Ethical and Legal Landscapes
Emphasis on explicit, informed consent and the right to withdraw consent.
Mandatory transparency and labelling of AI-generated content.
Clearer rules and liabilities for creators, distributors and platforms.
Increasing importance of public education and media literacy.
What you can do:
Be a digital superhero
For everyone
Strengthening digital competence: Learn how to spot fakes, question online content and understand deepfake technology.
Advocate for stronger laws: Support policies that improve platform accountability and protect individuals.
Think before you share: Check content before sharing it to be part of the solution.
For victims
Document everything: Screenshots, URLs, screen recordings with timestamps, accompanying texts/comments.
Report platforms: Use the integrated reporting procedures immediately.
Seeking legal assistance: Contact solicitors who specialise in cyberbullying and image-based abuse.
Contact law enforcement agencies: Especially in cases of serious damage, identity theft or CSAM (National Centre for Missing and Exploited Children (NCMEC) for child-related content).
Utilise tools for content removal: Stop NCII and Take It Down for digital fingerprinting and removal.
Conclusion: Protecting our reality in the age of AI
Deepfakes, particularly intimate AI-generated content, pose a significant threat to trust, privacy and dignity, and require urgent societal attention. Addressing this challenge requires a united front: individuals improving their digital literacy, technology companies developing ethical safeguards, and governments enacting strong, enforceable laws. The goal is to ensure that reality remains our own, built on truth, trust and respect, not fabricated misrepresentations and stolen identities.
Share This Article!
