The White House is launching a voluntary security review process specifically for closed-source AI models. This framework excludes developers who publish their underlying code. By targeting proprietary systems, the administration seeks to identify vulnerabilities before public release. This creates a distinct regulatory divide between open-source and closed-source AI development paths.