OpenAI is finalizing an advanced AI model featuring specialized cybersecurity capabilities. The company plans a limited release to a small group of selected firms.
This strategy mirrors Anthropic’s recent decision to restrict its Mythos model over malicious hacking concerns. AI developers increasingly worry about the dual-use nature of high-capacity models.
OpenAI aims to restrict access to ensure the tool supports defense and vulnerability discovery. Specific model capabilities and participating companies remain undisclosed.