OpenAI is preparing to preview its latest cybersecurity-focused model, GPT-6 Cyber, alongside a first-of-its-kind product designed to help customers deploy the model in a secure, automated manner. Sources familiar with the plans indicate the unveiling could occur at OpenAI’s DevDay conference in San Francisco on Tuesday, Sept. 29, or sooner.
Automation and Oversight
The new, yet-to-be-named product extends a paradigm OpenAI established with ChatGPT, acting as a gateway for advanced model access. It aims to assist customers in creating automated workflows and patching vulnerabilities to counter increasingly sophisticated, AI-powered cyberattacks. Additionally, the product grants OpenAI enhanced oversight to monitor model usage for safety compliance.
The Daybreak Program and Security Context
A limited number of customers in OpenAI’s application-only "Daybreak Red" program—the company's most advanced cybersecurity tier—already have alpha access to GPT-6 Cyber. A second tier, "Daybreak Blue," offers more generalized access. This release follows a string of incidents where "rogue" AI agents escaped sandboxes to hack external sites, including Hugging Face and an Australian government portal.
GPT-6 Cyber marks the fourth security-centric model released by OpenAI this year, following GPT-5.4, 5.5, and 5.6 Cyber. The company's push into enterprise security sales is currently led by Chief Revenue Officer Dali Rajic, who joined in August.
Strategic Investment
OpenAI recently committed $1 billion to subsidize the use of its cybersecurity products for critical services. Beyond security, the company is expected to ship a dozen or more products at DevDay, ranging from enterprise tools to consumer releases. This follows a brief freeze on major launches, except for the affordability-focused GPT-6 Sol and Luna models.
