08 May 2025

By Paul Haneke

As AI continuously enhances our everyday lives, for example by making Google a little more convenient and a lot less transparent, it is important not to lose sight of the threats it poses. Because as the technology evolves, so do they.

One of the most striking examples of unexpectedly rapid development made possible by generative AI, artificial intelligence that uses pattern recognition and machine learning to then produce something similar to what it was trained on, are deepfakes. These are visual representations of any existing person that have been manipulated in some way. And while just a few years ago, you could easily tell apart the “fake” from the “real”, this discernment has recently, within the span of just a few years, become scarily difficult, sometimes even impossible. It is used for a variety of content creation, some of it harmless, such as comedic and satirical purposes, but much of it is geared towards despicable exploitation and intentional misinformation, as for example voting interference by character assassination, revenge porn, and child pornography.

Tragically, and partly owing to the immense speed at which generative AI is currently advancing, the current state of deepfake regulation, whether it be on state or social media company level, is lacking. Things are certainly heading in the right direction with the German “Bundesrat” proposing stricter laws and higher criminal liability or the US governments proposal of the “Tools to Address Known Exploitation by Immobilizing Technological Deepfakes on Websites and Networks Act” or in short “TAKE IT DOWN Act”, which has already been passed by the House of Representatives and focuses on the issue of fighting revenge porn; things are just not moving fast enough.

“The TAKE IT DOWN Act” still has to pass President Trump without getting vetoed, which would not be out of character at all. The German government’s current means to pass laws are not looking all too shiny either. And while these laws take months if not years to be implemented, deepfakes and the tools to create them remain freely available to anyone with an internet connection and ugly intentions.

So, stay vigilant while venturing online and remember to triple check any “outrageous news story” that seems even a tiny little bit fishy, especially before sharing it and thus actively, albeit unknowingly, contributing to the spread of misinformation. Instead of an immediate “did your hear politician XY say this thing that was very out of character and evil?”, maybe a quick authenticity check through a different, more qualified medium, might serve as a healthier approach.

And remember that almost anything you see online could technically be AI generated.

Thanks, the article should work like that. You're welcome! Glad it fits the tone you're going for. If you need a title, subheading, or a punchy intro to match the ending, I’ve got you covered. Want help with that too?

(definitely real) sources and further reading:

https://www.weforum.org/stories/2025/01/deepfakes-different-threat-than-expected/

https://www.bpb.de/lernen/bewegtbild-und-politische-bildung/556843/strafrecht-und-regulierung-von-deepfake-pornografie/#footnote-target-7

https://news.mit.edu/2023/explained-generative-ai-1109

https://edition.cnn.com/2024/02/04/asia/deepfake-cfo-scam-hong-kong-intl-hnk/index.html

https://www.bundesrat.de/DE/plenum/bundesrat-kompakt/24/1046/15.html?nn=4352768#top-15