OpenAI launches cybersecurity model GPT-6 Cyber

on November 6, according to The Verge, OpenAI launched the cybersecurity model GPT-6 Cyber. The company stated that in tests simulating the previous Hugging Face AI agent intrusion event, the rate of boundary-crossing actions was less than 1%, lower than GPT-5.6 Cyber's approximately 5%.

In ExploitGym tests measuring models' vulnerability exploitation capabilities, GPT-6 Cyber scored 48.5%, higher than GPT-5.6 Cyber's 32.1% and GPT-6 Astra's 42.4%. OpenAI stated that the model helped discover 81 previously unknown vulnerabilities in Firefox.

OpenAI has classified the new model's cybersecurity capabilities at the 'Critical' level and added safety measures to detect and block abuse. The model is currently open to select members of the Daybreak Red program and will expand to all customers of the program over the coming weeks.