Cyber Command Builds Model-Agnostic AI Playbook
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- U.S. Cyber Command is building infrastructure to swap between AI models regardless of vendor or country of origin, with chief AI officer Brig. Gen. Reid Novotny saying 'no politics' will influence model selection — even for open-source Chinese models.
- Cyber Command received its first dedicated AI program funding in fiscal year 2026, which it's using to pilot commercial AI capabilities while building model-agnostic infrastructure underneath.
- Anthropic's Mythos Preview remains restricted to a patchwork of agencies — the NSA and the Commerce Department's AI testing institute have access, but CISA does not — as the White House continues negotiating.
- OpenAI is capitalizing on the gap by engaging federal, state, and international government offices to deploy its competing GPT-5.4-Cyber model, while Google has signed a classified deal with the Pentagon to use Gemini.
- Novotny confirmed fully autonomous AI deployment is off the table, saying Cyber Command tests varying levels of human oversight and would never unleash a 'human-out-of-the-loop tool' and then say 'Oops.'
- Lt. Gen. Lori Reynolds, former head of Marine Corps Forces Cyberspace Command, warned at Vanderbilt that AI will strain the military's 'persistent engagement' model built around a small number of sophisticated adversaries: 'It's insufficient by itself.'
Why it matters: By building vendor-agnostic infrastructure that can swap models on the fly, Cyber Command is decoupling its cyber operations from the political fight between Anthropic and the Pentagon — while OpenAI and Google sign classified deals that the Mythos standoff left open, giving them a durable foothold in the federal cyber AI market that Anthropic entered first.
Ask SkimNews
