This week’s Defence & Security AI news is dominated by a US policy inflection point: the Trump administration is actively dismantling Biden-era human-control guardrails for autonomous weapons while simultaneously accelerating AI procurement across every service branch. The tension between speed-to-deployment and meaningful human oversight is now a live operational and political crisis, with Congress, watchdogs, and allied nations all pushing back. Meanwhile, on the hardware front, the race to field AI-piloted autonomous combat aircraft reached a new public milestone at Eurosatory 2026.
Top story: Trump’s NSPM-11 orders the Pentagon to rewrite human-control rules for AI weapons within 90 days, triggering a Congressional backlash and setting the stage for the most significant shift in autonomous weapons policy in a decade.
Senator Demands Pentagon Explain Rushed Autonomous Weapons Policy Rewrite
DefenseScoop · Regulation
Following Trump’s signing of National Security Presidential Memorandum 11 on 5 June — which ordered the Pentagon to revise DoD Directive 3000.09, the foundational human-control policy for autonomous weapon systems, within 90 days — Senator Ruben Gallego wrote to Defence Secretary Hegseth demanding details on how the compressed timeline would protect US personnel and allies from unintended harm. The directive also mandates the military terminate contracts with AI companies that limit government use of their technology, widely seen as a direct response to the Anthropic standoff. For defence practitioners, this signals a fundamental rebalancing away from hard human-oversight requirements toward mission-speed autonomy, with accountability placed inside the chain of command rather than with external regulators.
Shield AI Unveils World’s First AI-Piloted VTOL Fighter at Eurosatory
Defence Blog · Generative AI
US defence tech firm Shield AI showcased the X-BAT at Eurosatory 2026 in Paris, billing it as the world’s first AI-piloted VTOL combat jet — designed to operate entirely without GPS or communications links, with a claimed range exceeding 3,700 km and first flights scheduled for later in 2026. The appearance at Europe’s premier land and air defence exhibition marks one of Shield AI’s most prominent European stages yet, and signals growing transatlantic appetite for fully autonomous air combat platforms. The X-BAT’s runway-free design directly challenges assumptions about fixed airbase dependency, a vulnerability exposed repeatedly in Ukraine and the Middle East.
GAO: US Navy Failing to Deliver on AI Autonomous Fleet Despite Urgent Need
Military Times · Strategy
A declassified GAO report published on 15 June found the US Navy has still not fully implemented recommendations made over a year ago to streamline its robotic and autonomous systems programme, despite publicly concurring with the advice. The watchdog found that compartmentalising investment by domain forces new autonomous technologies to compete against legacy weapons budgets, and urged the Navy to adopt platform-agnostic, iteratively developed capabilities — the commercial-sector model used by companies like SpaceX. With the Navy having stood up a new Portfolio Acquisition Executive for RAS in December 2025 to consolidate 47 disparate programmes, the report is a critical stress-test of whether that reorganisation will actually close the delivery gap exposed by drone warfare in Ukraine and the Middle East.
US Army Issues RFI for Agentic AI Deployable on Top-Secret Networks
ExecutiveGov · Tools
The US Army has issued a formal Request for Information for agentic AI platforms capable of operating on its most sensitive classified networks — including JWICS and SIPRNet at Impact Level 6 — enabling intelligence analysts to build, deploy, and oversee AI agents that autonomously retrieve information and execute tasks in classified environments. The platform must also function in edge and disconnected or bandwidth-limited settings, anticipating contested battlefield conditions. This represents a significant maturation in military AI procurement: moving beyond large language model assistants toward autonomous agent frameworks embedded directly in the operational intelligence stack.
EU Commission Welcomes G7 Cyber Declaration as Tech Sovereignty Package Lands
European Commission (digital-strategy.ec.europa.eu) · Regulation
On 8 June, the European Commission welcomed the G7 Cybersecurity Working Group Declaration, framing collective cyber defence against evolving digital threats as a strategic priority — coming just days after the Commission tabled its European Technological Sovereignty Package on 3 June, a suite of measures to strengthen Europe’s own capacity in semiconductors, AI, and cloud infrastructure. Together, the two announcements mark a deliberate European push to reduce dependence on non-EU AI and cloud providers for security-critical systems, directly relevant to the ongoing debate about US-vendor lock-in in NATO defence infrastructure. For UK and European security practitioners, this signals tightening procurement and supply-chain expectations around sovereign AI capability.