Deepfakes are marvels of technology that allow Luke Skywalker to appear young in the Boba Fett series. We can put any face on any person’s body, and make that face move naturally, more and more.
The problem is that it also allows the digital identity of people to be supplanted without their permission, which is fertile ground for Fake News.
The point is that Google has now banned the training of AI systems that can be used to generate deepfakes on its Google Collaboratory platform. From now on, in the list of not allowed projects, we have deepfakes, since apparently it was being used for purposes
Google Collaboratory, also called Colab, was born in 2017 as an internal project of Google Research, and allows anyone to write and run Python code from the browser, for machine learning, education and data analysis. In this way, anyone can have access to GPUs and processing units that are ideal for projects related to Artificial Intelligence.
There is an open source deepfake generator called DeepFaceLab, and before you could use Colab to do your projects. Now the message appears:
You may be running code that is not allowed and this may restrict your ability to use Colab in the future. Please note the prohibited actions specified in our FAQ.
Google told TC that the block is due to regular abusive pattern reviews they do from time to time, blocks that are in addition to running denial-of-service attacks, cracking passwords, and downloading torrents.
The decision is not final, as Google acknowledges that deepfakes can be very beneficial, but have great potential to go against Google’s AI principles. If you want to use it for good, you’ll need to speak with a Google Cloud representative to review each use case.