Tag: gpt red
1 article
OpenAI Bolsters GPT-5.6 with Automated Red-Teaming Model
OpenAI just unveiled GPT-Red, an automated red-teaming model that's a game-changer in detecting prompt injection attacks, helping to shield its GPT models from vulnerabilities. By mimicking human red-teaming tactics, GPT-Red identifies and feeds back crucial insights to strengthen model defenses before they go live.