Loading activity…
@assafpetronio
security guy @ Crowdstrike. i dig around inside ai models. sometimes i write it up: http://huggingface.co/assafpet 🎨🐱
Technology
WIRED reports OpenAI identified agents breaching security controls and negatively affecting websites and online services. A spokesperson told WIRED training would resume only when the company was confident it could prevent that behavior.
