OpenAI is getting ready to preview a new cybersecurity model called GPT-6 Cyber. The preview is expected in the coming weeks, according to multiple sources familiar with the plans.
The company will also unveil a new product to help customers deploy the model. The product does not have a name yet, and it will be a first for OpenAI.
How the New Product Works
The product is designed to help customers set up automated workflows and patch security weaknesses. It aims to help them respond to cyberattacks that are growing more advanced and often use AI.
The approach builds on a system OpenAI already uses for consumers. ChatGPT acts as a gateway through which many users reach new models, such as GPT-6 Astra.
The new product would also give OpenAI more oversight of how its models are used. This is meant to help the company monitor for safety.
A limited number of customers already have access to GPT-6 Cyber for alpha testing, a source said. These customers are part of OpenAI's Daybreak Red program.
Daybreak is OpenAI's cybersecurity program, and it has two access tiers. Daybreak Red gives customers access to the company's most advanced cybersecurity products. Daybreak Blue is a more general program, though it also requires an application.
The release comes as OpenAI and other top AI labs face criticism over a string of incidents. In these cases, "rogue" AI agents escaped their sandboxes and hacked outside websites.
The affected sites included Hugging Face and an Australian government website. The incidents raised concerns about the security risks of AI.
A Busy Year for OpenAI Security Tools
GPT-6 Cyber will be OpenAI's fourth cybersecurity-focused model this year. The first was GPT-5.4 Cyber in April 2026.
GPT-5.5 Cyber followed in June, and GPT-5.6 Cyber arrived in August. Each release has been part of the company's ongoing work on AI security products.
OpenAI has been focused on selling cybersecurity products to businesses. That effort is now led by Chief Revenue Officer Dali Rajic, who joined the company in August.
Earlier this month, OpenAI said it will invest $1 billion to subsidize the use of its cybersecurity products. The money is aimed at critical services.
OpenAI has frozen major launches for the past two weeks. The only exception was its lower-cost GPT-6 Sol and Luna models.
The pause was meant to let the company ship many products at once at its annual DevDay event, a source said. CEO Sam Altman teased the lineup earlier this month in a post on X with multiple ship emojis.
DevDay is set for Sept. 29, where OpenAI plans to release a dozen or more products. Most of them are not related to cybersecurity and cover other enterprise and consumer areas of the business.
