Google Quietly Bans Deepfake Training Projects on Colab (bleepingcomputer.com) 5
Google has quietly banned deepfake projects on its Colaboratory (Colab) service, putting an end to the large-scale utilization of the platform's resources for this purpose. From a report: Colab is an online computing resource that allows researchers to run Python code directly through the browser while using free computing resources, including GPUs, to power their projects. Due to the multi-core nature of GPUs, Colab is ideal for training machine learning projects like deepfake models or for performing data analysis. Deepfakes can be trained to swap faces on video clips, adding realistic facial expressions to make the result appear genuine, although it's fake. They have been used for spreading fake news, creating revenge porn, or for fun. However, the lack of ethical limitations in their use has been the source of controversy and concern.
They don't want competition (Score:3, Interesting)
Seems fair, read the details (Score:5, Informative)
Google Colab was never intended to train your production-ready datasets. It's for using spare resources to learn how to do ML. And in particular DeepFaceLab has a tutorial on using your GCP free credits to train a dataset, just copying and pasting code, letting it spin on TPUs, and then collecting the output.
Seems like a fine enforcement action to me. I've benefited from their free TPU access, likely saving me hundreds of dollars as I learned ML shit with GPT-2 and GPT-J 6B fine-tuning, which I can't do on my home GPUs.
Yo Dawg (Score:3)
I heard you like AI
So I made an AI to police how you use AI
Why not? (Score:2)
Well, the proper Subject is "Why no interest on Slashdot in this particular abuse of technology?" It's at the bottom of the top page with a total of three comments...
Looked a the few comments for hints, but didn't find any. So I guess I should ask why I didn't find the story compelling?
I am sure (Score:2)
I am sure crypto mining is not allowed either. Nor any other abuse of free resources.
Yes, it is fun to train deep fake models. However given their complexity, and popularity, I would expect them to be on the very top end of resources used.