#robustness

OpenAI Introduces GPT-Red: A Self-Improving Safety Model for Robustness Testing

OpenAI has published GPT-Red, a safety research model that autonomously probes and hardens other AI systems through self-improvement loops and adversarial testing.