OpenAI unveils a new cyber model as AI attacks increase

Reports of AI agents threatening cybersecurity and committing various violations have been increasing recently. To prevent such dangerous incidents, leading technology companies are expanding their protective tools. According to ixbt.com, OpenAI announced that it had expanded its Daybreak cybersecurity service and developed a specialized protection model. TechCrunch.com reports this.

The next stage of the Daybreak service

The updated Daybreak service presented by OpenAI is now divided into two main categories: Blue and Red. Both categories give verified customers access to advanced cyber models. Previously, strict restrictions and security barriers were in place for using such cutting-edge technologies.

The first tier, called Blue, serves as the recommended starting point for most businesses. The service provides various security capabilities, including incident response, malware analysis, and software update checks.

The GPT-5.6-Cyber model and its capabilities

The Red category includes a broader and more sophisticated set of tools. It provides users with specially trained cybersecurity models for security testing and vulnerability research. The new GPT-5.6-Cyber model was also launched within this category.

According to the company, the GPT-5.6-Cyber model was built on GPT-5.6 Sol and offers expanded capabilities for performing certain specialized cybersecurity tasks. At present, this advanced model is available only to trusted partner customers such as Accenture, IBM, Crowdstrike, and Cloudflare.

Experts and critics say that while threats from AI agents are growing rapidly, they also provide major AI laboratories with an effective marketing opportunity. OpenAI is presenting its updated service in precisely this spirit.

In its blog, the company emphasized that the cybersecurity landscape is changing rapidly and that threat actors are increasingly using AI to carry out cyberattacks at unprecedented speed and with full autonomy. As these capabilities expand, defenders have less time to prepare.