Source: Mixed News
Google has reportedly banned its deepfake projects on its Colaboratory, ‘Colab’ service quietly. Mainly, this officially suspends the large scale utilisation of the site’s resources for this particular purpose. Essentially, Colab is an online computing resource that mainly enables researchers to directly run Python code through the browser. This includes allowing to run it through the browser while they use free computing resources such as GPU’s to power their projects.
Owing to the GPUs’ multi core nature, Google’s Colab is perfect for training machine learning such as deepfake models or for performances of data analysis. One can train deepfakes to swap people’s faces on video clips, along with training it to add realistic facial expressions on them to make them appear more authentic. However, one should keep in mind that all of these changes would be fake.
Deepfakes have been used by entities to spread fake news, make ‘revenge porn,’ or simply for fun which might prove to be harmful to others. Clearly, the absence of ethical limitation in its use had led to a significant range of controversy and concern over its existence.
The ban on deepfakes:
According to archive.org historical data, the ban has been in place since early May. Rather quietly, Google Research simply added deep fakes to its list of disallowed projects. DFL developer, ‘chervonij’ took to Discord to let people know that deepfakes is no longer allowed on Colab. Whoever tries to train deepfakes on the platform of Colab, they would be served with an error message telling them ‘executing a disallowed code’ could affect their use of Colab in the future.
Many expect the affect of this new ban to be relatively far-reaching in the world of deepfake as many utilised pre-trained models with Colab to kick off their high resolution projects. All this while, the platform was making the process rather simple even for ones who had no previous knowledge of coding. Clearly, this is why several tutorials suggest Google’s “free resource” outlet to launch such projects.
However, it is unsure if the platform performed the policy owing to ethical reasons, or the frequent use of the free computing resources for such project. Ideally, Colab is supposed to be used by researchers needing power that several thousands dollars to aid them in reaching scientific goals, especially at times of GPU shortages. Unfortunately, various reports suggested that certain users were abusing the free tier of the platform for the creation of deepfake models at scale.
