[OpenAI Releases Cyber Model to Select Users, Fine-Tuned for Cybersecurity Use Cases] Foresight News reports that OpenAI has announced the expansion of its cybersecurity trusted access features. Top-tier clients can apply for access to GPT-5.4-Cyber, a version of GPT-5.4 fine-tuned specifically for cybersecurity use cases, enabling more advanced defense workflows. A week ago, Anthropic PBC announced the limited release of the AI tool Mythos.