OpenAI Launches GPT-5.5-Cyber Model for Vetted Security Researchers
⢠OpenAI released GPT-5.5-Cyber, a specialized AI model with relaxed restrictions accessible only to vetted security researchers protecting critical infrastructure. ⢠The model supports cybersecurity tasks like malware analysis and authorized penetration testing, overcoming blocks in standard chatbots, and features a three-tiered access system from public to least restrictive Cyber variant. ⢠This positions OpenAI in competition with Anthropic's Mythos Preview, aiming to empower AI defenders amid industry concerns over offensive AI capabilities.
agentupdate.ai