This week’s defence AI landscape is defined by an accelerating data-and-dollars arms race: record venture capital is pouring into autonomous weapons startups, Ukraine’s battlefield footage is being industrialised as AI training fuel, and Western legislators on both sides of the Atlantic are scrambling to put governance guardrails around systems that are already being deployed. The gap between what AI can do on the battlefield and what rules exist to govern it has never been wider.
Top story: Over 500,000 hours of real Ukraine combat drone footage are now available to train military AI targeting models — a watershed moment for autonomous weapons development.
Ukraine’s Half-Million Hours of Combat Footage Now Trains Military AI
TechRadar · Generative AI
Virginia-based startup Enabled Intelligence has released over 500,000 hours of pre-labelled drone footage from the Ukraine conflict into its EView library, making it available to defence contractors and NATO-allied nations to train autonomous targeting AI. The footage covers aerial object detection, vehicle classification, and ground activity across real weather and terrain conditions that no simulation can replicate — representing a fundamental shift in how military AI models are built. As one analyst put it, dataset quality is now a bigger differentiator in autonomous warfare than the quality of the drone hardware itself.
FY27 NDAA Advances New AI Mandates for All US Defence Contractors
GovInfoSecurity · Regulation
Both the House and Senate Armed Services Committees have advanced separate versions of the FY2027 National Defence Authorization Act, each authorising $1.15 trillion for national defence and imposing significant new AI security obligations on the entire defence industrial base. Defence contractors would face tighter CMMC cybersecurity certification requirements alongside new AI-specific guardrails governing how Pentagon AI is developed, deployed, and monitored on military networks. The dual-chamber move signals that AI governance — not just AI procurement — is becoming a core compliance requirement for any company doing business with the US military.
https://www.govinfosecurity.com/ndaa-drafts-bolster-contractor-cyber-ai-rules-a-32029
Defence AI Startup Funding Shatters Records at $14.6bn in Five Months
TechTimes · Finance
More than $14.6 billion flowed into military, national security, and law enforcement AI startups in just the first five months of 2026 — already surpassing the previous full-year record of $9.6 billion set in 2025, according to Crunchbase data. The surge is concentrated in a handful of platform companies: Anduril ($5bn Series H at a $61bn valuation), Shield AI ($1.5bn Series G), and Saronic ($600m). This investment explosion is happening simultaneously with a mandatory 90-day Pentagon rewrite of the legal rules governing what AI-powered, machine-speed kill chains may actually do — meaning capital is outpacing governance by a significant margin.
IDF Expands AI-Targeting SMASH Hopper to Dozens of New Platforms
All Israel News · Strategy
The Israel Defence Forces has procured an additional NIS 7.4 million ($2.5m) in spare parts and services for the SMASH Hopper, an AI-assisted remote-controlled weapon station from Smart Shooter, enabling the system to be mounted on dozens of additional platforms across the country. The move reflects the IDF’s deepening reliance on AI in post-Iran-war operations, as AI-enabled systems are increasingly integrated into layered defences against missiles, rockets, and drones. Smart Shooter has also secured contracts to supply its systems to the US, UK, and Germany — signalling that combat-proven Israeli AI targeting technology is rapidly entering allied military supply chains.
https://allisraelnews.com/idf-takes-another-step-into-the-artificial-intelligence-revolution