AI moves deeper into defence

Artificial intelligence is moving from back-office military software into war planning, battlefield analysis and the politics of who gets access to the technology. (bloomberg.com) (mod.gov.ua) (benzinga.com) At the Pentagon, Google’s Gemini for Government is already being used on unclassified networks through GenAI.mil, a Defense Department platform that Google said reaches more than 3 million civilian and military personnel. Bloomberg reported on March 10 that the first rollout was for routine work on unclassified systems. (cloud.google.com) (bloomberg.com) That matters because a reported move into classified operations would be a step beyond drafting emails or summarizing documents. Google’s public announcements so far have described accredited government cloud services and unclassified use, not classified combat or intelligence workflows. (cloud.google.com 1) (cloud.google.com 2) Ukraine is making a parallel push from the battlefield outward. Its Ministry of Defence said last month that it is launching the Defense AI Center “A1” as its first center of excellence for integrating artificial intelligence into defense processes and warfare technology. (mod.gov.ua) (kmu.gov.ua) The ministry said A1 will work on battlefield analysis, data processing, command support and partnerships that let companies test systems in Ukraine. It also said the project is being implemented with support from the UK government. (mod.gov.ua) (interfax.com.ua) Artificial intelligence in this context usually means software that sorts huge volumes of text, images, video or sensor data faster than human analysts can. In military use, that can mean flagging patterns in drone footage, helping staff search documents, or generating options for planners before a human signs off. (cloud.google.com) (mod.gov.ua) The commercial fight around those systems is now colliding with national security policy. Nvidia Chief Executive Jensen Huang said this week that treating China only as an adversary in artificial intelligence could backfire, warning that cutting China off could weaken the U.S. technology stack’s global reach. (benzinga.com) (blogs.nvidia.com) Huang’s position cuts against the logic behind tighter chip controls, which aim to slow China’s access to advanced computing. His argument is that U.S. companies also need China’s developers and market if they want American software and hardware to remain the default worldwide. (benzinga.com) (blogs.nvidia.com) Taken together, the three developments show the same shift at different levels: the Pentagon is scaling enterprise AI, Ukraine is wiring AI into wartime operations, and Nvidia is lobbying over the global rules of the system. The next fights are likely to center on where these models can run, what data they can see and how much autonomy militaries will allow them. (bloomberg.com) (mod.gov.ua) (benzinga.com)