In the coming weeks, OpenAI plans to release a preview version of its latest cybersecurity model, GPT-6 Cyber. The company also announced a new product that will help customers deploy the model in a more secure and automated way, according to Fortune magazine, citing people familiar with the matter.
Image source: AI
The new, as-yet-unnamed product will be OpenAI’s first of its kind and will continue a strategy the company has established with consumer products: ChatGPT acts as a gateway through which many users can access new models such as GPT-6 Astra. The solution is designed to help customers create automated workflows and reduce vulnerabilities in the face of increasingly sophisticated artificial intelligence cyberattacks. It will also give OpenAI more control over how its models are used for security purposes.
One of the sources said that a limited number of customers participating in the Daybreak Red program (available by application only) have been given access to GPT-6 Cyber for alpha testing. Daybreak is OpenAI’s cybersecurity program and is divided into two access levels: red and blue. Daybreak Red members have access to the company’s most advanced cybersecurity products. Daybreak Blue is a more broadly accessible program, but participation in the program is only available by prior application.
The launch of the new cybersecurity product comes as OpenAI and other leading artificial intelligence labs have been criticized for a series of disturbing incidents in which autonomous agents broke out of sandboxes and hacked external network resources, including the Hugging Face website and an Australian government website. While these cases raise serious concerns about security risks associated with artificial intelligence, OpenAI’s new tools are part of the company’s ongoing efforts to create secure products in the field.
GPT-6 Cyber is the fourth network security-focused model released by OpenAI this year. The first is GPT-5.4 Cyber (released in April 2026), followed by GPT-5.5 Cyber (released in June) and GPT-5.6 Cyber (released in August).
OpenAI has a strong focus on selling cybersecurity products to enterprise customers. The direction is led by commercial director Dali Rajic, who joined the company in August. Earlier this month, OpenAI announced it would invest $1 billion to subsidize the use of its cybersecurity products in critical infrastructure.
OpenAI also plans to unveil about a dozen new products next week as part of its annual DevDay conference scheduled for September 29, according to another source familiar with the event. Most of them are not related to cybersecurity, but to other areas of company activity, targeting enterprises and the mass market.
Sources say that over the past two weeks, OpenAI has paused all major releases — except for the accessibility-focused GPT-6 Sol and Luna models — in order to show them all off at DevDay. The company’s chief executive, Sam Altman, also hinted at a number of new products coming earlier this month.
If you find an error, select it with your mouse and press CTRL+ENTER.
