Anthropic and OpenAI voluntarily restrict distribution of their most advanced AI models specialized in cybersecurity, fearing misuse or unauthorized copying.
This limitation reflects tensions between commercial accessibility and security risks associated with cutting-edge generative AI technologies.