Software can strip AI safety measures from Meta, Google models in minutes

Software tools that remove safety protections from AI models developed by Meta, Google and other tech groups are being used to create thousands of altered versions stripped of their original controls. Read More