OpenAI has introduced GPT-6 Cyber, a specialised artificial intelligence model designed to support authorised cybersecurity work, including vulnerability discovery and security testing.
According to the system card published on November 5, the model is a cybersecurity-focused variant of GPT-6 Astra and has undergone specialised training to improve its capabilities in security-related tasks.
The company said the system is intended to help authorised security professionals identify weaknesses in digital systems and strengthen their defences.
OpenAI classified the model as having critical cybersecurity capabilities under its Preparedness Framework, while maintaining that access would be restricted.
The model is being made available through Daybreak Red, a controlled access programme requiring identity verification and organisational security measures.
These include single sign-on, multi-factor authentication and restrictions on the environments in which the system can be used.
The development comes as organisations face increasingly complex cybersecurity challenges and technology companies explore how advanced AI systems can support defensive security work.
However, the model’s capabilities also underline the importance of safeguards to prevent powerful cybersecurity tools from being used outside authorised environments.
