This week’s defence AI landscape is defined by a tug-of-war between acceleration and accountability: defence tech funding smashed records while multiple US Senate bills push hard for human-in-the-loop mandates on autonomous weapons. Simultaneously, Europe made its boldest sovereignty play yet by backing its own open-source frontier AI model, signalling that strategic autonomy — not just regulation — is now Brussels’ priority.
Top story: US Senate HALO Act would legally require a named human commander to authorise every AI-assisted lethal engagement, directly challenging how systems like Anduril’s Lattice and Shield AI’s Hivemind are currently built to operate.
Senate HALO Act Would Ban AI From Making Kill Decisions Alone
The Debrief · Regulation
Senator Adam Schiff introduced the Human Authority in Lethal Operations (HALO) Act on June 8, which would impose sweeping design standards and legal safeguards requiring autonomous and semi-autonomous weapon systems to remain under meaningful human control at all times. The bill mandates that a designated commander retain ‘ultimate discretion, judgment, and control over the use of force’, and would require existing non-compliant systems to be reviewed, modified, or retired. It arrives as the Pentagon aggressively scales autonomous systems, creating a direct legislative collision with the trajectory of US defence AI procurement.
Defence Tech Funding Hits $14.6Bn in Five Months — A New Record
TechTimes · Finance
Defence tech investment reached $14.6 billion in just the first five months of 2026, already exceeding the previous full-year record of $9.6 billion set in 2025, driven by Anduril Industries’ $5 billion Series H round that valued the company at $61 billion. The Army has also awarded Anduril a 10-year, up to $20 billion enterprise contract to deploy its Lattice counter-drone platform. The concentration of capital into a small number of platform companies signals a maturing of the defence AI sector, with institutional investors replacing early-stage venture funds as the primary backers.
Senate FY27 NDAA Bans AI From Launching Nuclear Weapons Autonomously
Crypto Briefing · Regulation
The Senate’s FY2027 National Defense Authorization Act now includes provisions explicitly prohibiting AI from autonomously launching nuclear weapons or employing lethal force without a human authorising the decision, driven by Senators Slotkin and Gillibrand. The amendments represent the most concrete legislative effort yet to draw hard legal lines around Pentagon AI deployment in life-or-death scenarios, complementing the HALO Act introduced the same week. For defence contractors and AI firms supplying DoD, these overlapping bills are converging into a regulatory framework that could fundamentally reshape system architecture requirements.
https://cryptobriefing.com/senate-ndaa-bans-ai-nuclear-weapons-lethal-force/
EU Picks EUROPA Consortium to Build Sovereign Frontier AI Model
European Commission (digital-strategy.ec.europa.eu) · Strategy
The European Commission selected the EUROPA consortium, led by Italian firm Domyn in collaboration with Fraunhofer-Gesellschaft, to build an open-source frontier AI model covering all 24 official EU languages as part of its Frontier AI Grand Challenge. The model will exceed 400 billion parameters and run entirely on European supercomputing infrastructure, reducing dependence on US and Chinese AI systems. For the defence sector, the move matters strategically: a permissively licensed, EU-governed model gives European defence institutions and regulated industries a deployment option that sidesteps both vendor lock-in and data-residency risks associated with American proprietary APIs.
Ukraine Confirms First Fully Autonomous AI Drone Kills on Battlefield
Wikipedia / New Scientist (via Wikipedia) · Risk
A senior figure in Ukraine’s defence industry disclosed in June 2026 that fully autonomous drones with no human oversight killed soldiers on the battlefield for the first time two years ago, involving ten AI-controlled ‘Terminator’ drones that killed Russian soldiers and targeted a tank with no operator connection. The disclosure is significant because it is the first confirmed public acknowledgement that lethal fully autonomous AI weapon systems have already been used in combat, not as a future scenario. The revelation is directly fuelling the wave of US Senate legislation demanding human-in-the-loop safeguards, and will intensify international pressure for binding treaty controls on autonomous weapons.