Red Hat drops LibreOffice, Mozilla AI contest winners, Reddit's Strike, and more!
Mozilla awards 100.000$ in “Responsible AI” prizes In March, Mozilla launched the “Responsible AI Channel”, a one-day in-person event to build “trustworthy AI products”. This is extremely important because AI tools currently raise privacy and copyright concerns. So, who won? The top prize, $50k, was awarded to the “Sanative AI” project. The idea here is to “watermark” images through a small amount of visual noise that will make them useless upon training (as “the AI is unable to properly minimize its loss function”). Creating this watermark, which is different for each image, requires a great number of resources, and yet Sanative offers it for free through a queue. This approach has some potential issues: already uploaded images - without this watermark - can still be used for training, and AI organizations might find a way to get around the noise. However, Sanative AI is committed to improving its tool over time to remain effective. ...