OpenAI’s new security model is for “critical cyber defenders” only

OpenAI is preparing to launch a new cybersecurity model, GPT-5.5-Cyber. CEO Sam Altman said the model will not be available to the general public, but will be rolled out first to a select group of trusted “cyber defenders” so organizations can shore up their cyber defenses.

The limited rollout will take place “in the next few days.” Altman said in X. “We will work with the entire ecosystem and government to create reliable internet access.”

However, it is not clear who will be able to access the model first former “Trusted access” Charts Involve vetted professionals and institutions. The details of the model and its capabilities are not clear; OpenAI has not released any technical details or specifications. The name indicates that it is a specialized version of Recently released GPT-5.5which she described as “the smartest and most user-friendly model yet.”

The rolling rollout is part of a growing trend in the AI ​​industry of companies describing their top models as too dangerous to go public due to the potential for abuse. OpenAI has released previous models focused on cybersecurity, as well as its new models designed for this purpose GPT-Rosalind Life Sciences Modelwhich aims to support biology research and drug discovery. This month, Anthropic followed a similar playbook with Cloud Mythos, but with much more fanfare. Secure form release failed In embarrassing ways.

The White House has shown keen interest in launching Mythos, despite ongoing tensions with Anthropic following its battle with the Pentagon. She recently opposed plans to expand access to Mythos further, According to to The Wall Street Journal. White House officials who were not named in the report cite cybersecurity concerns associated with a greater number of people with access to Mythos and concerns that increased demand could hamper the government’s ability to use the system as reasons to respond.