OpenAI Unveils GPT-Red to Test AI Model Safety

While red teaming is standard practice, using humans and AI to test the security of new models is novel. Enterprises should still ensure the model they use aligns with their business and security workflows.