Safety
GPT-Red: Unlocking Self-Improvement for Robustness
HumanOpenAI’s GPT-Red is an automated red-teaming model trained via self-play; GPT-5.6 Sol shows 6× fewer hard prompt-injection failures.
Knowledge catalogue
OpenAI expands Daybreak into Blue/Red tiers and introduces GPT-5.6-Cyber for authorized defensive vulnerability research.