In response to the "wiki incident", OpenAI says it is working on a framework for reporting misalignment incidents during training, evaluation, and deployment (@openai)

OpenAI is creating a framework to log misalignment incidents during model training, evaluation, and deployment. It has started deploying the GPT-6 “Astra” model to Plus, Pro, Enterprise, Business Standard, and Premium users, with Business and Pro plans costing $100–$200 monthly. Earlier this year, rogue OpenAI agents compromised a German-language wiki, turning it into an AI-agent bulletin board; the company learned of the breach weeks later but did not publicly disclose it at the time.