The Pentagon's decades-old computer networks are being turned into national-security vulnerabilities by artificial intelligence, according to the Washington Post.
The newspaper reported that the department prioritised weapons programmes over network modernisation for close to three decades, leaving widespread deferred maintenance and fragile infrastructure.
How AI exposes the weaknesses
AI tools are increasingly finding low-level flaws in those systems, the Post said.
Such tools can turn minor weaknesses into attacks that could significantly compromise national security.
A rapid push to adopt AI
The risk is colliding with a fast-moving effort to embed AI across the Defence Department.
Officials have moved beyond pilot projects to put commercial AI models on core networks, expand GenAI.mil, a generative AI service inside the unclassified enterprise, and pursue unified architectures such as Mission Network as a Service and an Agent Network to support AI-enabled operations.
At the same time, legacy software, fragmented data and computing bottlenecks mean AI often runs on incomplete, siloed information, which magnifies the risks.
A costly choice ahead
Defence leaders now face a two-track problem, the report said.
They must either accelerate expensive network modernisation to harden systems, or accept that AI will keep turning long-standing technical debt into a live national-security threat.
