Agencies proactively share with other agencies the custom-developed code, models and model weights of AI in active use, whether built or bought, and where practicable release and maintain the code as open source in a public repository, unless the law or IP, export control or classification rules forbid it, sharing would identifiably endanger national security, government confidentiality, individual privacy, or public rights and safety, a contract prevents it, or sharing would endanger agency missions, operations or systems. The datasets behind AI development and testing go into the agency's comprehensive data inventory where the OPEN Government Data Act and M-25-05 require, and agencies assess the risk that a model discloses sensitive details of its training data.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.