Quietly, Google has banned the Deepfake project on Colab

It is reported that Google has quietly banned its deepfake project on the Colaboratory (Colab) service recently, which represents the large-scale use of platform resources for the purpose of Deepfake. The era may have come to an end.

As we all know, Colab is an online computing resource platform that allows researchers to run Python code directly through the browser while using free computing resources, including GPUs, to support their projects. Due to the multi-core nature of GPUs, Colab is an ideal choice for machine learning projects like Deepfake models or for performing data analysis.

After certain training, people use Deepfake technology to swap faces in video clips and add real facial expressions, almost making the fake look real. However, the technology is often used to spread fake news, create revenge porn, or for entertainment purposes. The lack of ethical restrictions in practical applications has been a source of controversy for this technology.

Deepfake is banned

According to historical data from the Internet Archive website archive.org, this ban was issued earlier this month. Google Research quietly included Deepfake as a banned project. in the list.

As DFL software developer "chervonij" pointed out on the Discord community platform, users who are still trying to train deepfake on the Colab platform will receive this error report:

"You may be executing code that is not allowed, which may limit your ability to use Colab in the future. For more information about prohibited behavior, please pay attention to our Frequently Asked Questions (FAQ) document."

Analysts expect that this new restriction will have a very profound impact in the deepfake world, because many users are currently using Colab’s pre-trained models to start their high-resolution projects. Even for those without a coding background, Colab can make the project process very smooth, which is why so many tutorials recommend users to use Google's "free resources" platform to start their own Deepfake projects.

Abuse of Resources

It is unclear whether Google implemented the ban out of ethical considerations or due to the misuse of free computing resources used by the project.

Colab was founded to help researchers who need thousands of resources to achieve scientific goals, which is particularly important in this era of GPU shortage.

However, it has backfired. Reports show that some users are using the platform’s free resources to create Deepfake models on a large scale, which has occupied a large amount of Colab’s available resources for a long time.

The following is the complete list of items that are not allowed:

• File hosting, media services or other web service products not related to Colab interactive computing
• Download torrents Or perform peer-to-peer file sharing
• Use a remote desktop or SSH service
• Connect to a remote proxy
• Cryptocurrency mining
• Run a denial of service attack
• Password cracking
• Use multiple accounts to resolve access or resource usage restrictions
• Create Deepfakes

It is not difficult to see that all these banned projects are far from Not eligible to conduct routine scientific research. While some projects fall within the scope of fair use, Google has found far more abuse than fair use.

Reference source: https://www.bleepingcomputer.com/news/technology/google-quietly-bans-deepfake-training-projects-on-colab/

The above is the detailed content of Quietly, Google has banned the Deepfake project on Colab. For more information, please follow other related articles on the PHP Chinese website!