The agency is discussing the matter with developers under White House directives
The National Security Agency wants access to “all” AI models, deputy director Tim Kosiba said on 27 August, according to a NextGov/FCW report. Speaking at an event in Bethesda, Maryland, Kosiba said: “We want access to all models.”
According to the report, the NSA is implementing White House AI directives and discussing the matter with leading developers. Kosiba did not identify the companies involved, the models used by the NSA or any access to unpublished models. Access could involve using a model through a company-controlled service, installing it in a government environment or accessing it for security testing on classified networks.
An executive order issued in June assigns the NSA responsibility for determining which AI systems qualify as “covered frontier models,” according to the report. Developers can voluntarily join an agreement that allows the federal government to access models for 30 days before their release to examine potential hacking capabilities. Another June directive calls on intelligence and defence agencies to form partnerships with the AI industry, make the most advanced models available to staff and avoid dependence on a single company.
