This week’s defence AI landscape is defined by a race between capability and accountability: AI is moving from experimentation to live deployment across weapons systems, cybersecurity, and intelligence — but lawmakers on both sides of the Atlantic are scrambling to define who, or what, holds the trigger. The Farnborough Airshow crystallised this tension, as autonomous combat aircraft were unveiled at the same moment the US Congress was war-gaming AI-enabled attacks on critical infrastructure.
Top story: BAE Systems unveiled Britain’s first sovereign AI-powered collaborative combat aircraft, ‘Brontanax’, at Farnborough — as the wider airshow confirmed autonomous AI weapons systems are now the central story in Western defence procurement.
BAE Systems Unveils Britain’s First Autonomous AI Combat Aircraft at Farnborough
Aerospace Global News · Strategy
BAE Systems debuted ‘Brontanax’, Britain’s first sovereign collaborative combat aircraft, at the 2026 Farnborough International Airshow. Roughly the size of a Hawk trainer, the uncrewed jet is designed to fly alongside Typhoon, F-35 and the future Tempest, delivering electronic warfare and precision-strike capabilities. The unveiling came as autonomous combat drones from Boeing, Anduril, General Atomics and Airbus made Farnborough the most defence-contractor-heavy airshow in memory, signalling that AI-piloted weapons are no longer a future concept but a procurement reality.
https://aerospaceglobalnews.com/event-wrap/farnborough-airshow-2026-day-three-wrap/
UK’s £5bn AI Drone Plan Quietly Crosses the Human-Control Threshold
House of Commons Library · Regulation
The UK’s Defence Investment Plan, published on 30 June 2026, commits more than £5 billion to drones and autonomous systems — the largest such commitment in British military history — and includes language indicating that some future weapons will be designed to make targeting decisions without a human authorising each individual strike. The plan includes Project NYX, a British Army programme to field up to 24 armed autonomous drones by 2030, alongside AI-enabled ‘loyal wingman’ aircraft for the RAF. For defence practitioners, this represents a formal policy shift away from the long-standing principle that a human must authorise every use of lethal force.
https://commonslibrary.parliament.uk/research-briefings/cbp-10935/
World’s First Fully Autonomous AI Ransomware Attack Documented by Researchers
Dark Reading · Risk
Sysdig’s Threat Research Team has documented JadePuffer, the first end-to-end ransomware operation executed autonomously by a large language model, with no human operator involved during the attack itself. The AI agent handled reconnaissance, credential theft, lateral movement, privilege escalation, database encryption and even wrote its own ransom note — adapting to obstacles in real time. For security and defence organisations, the finding signals a qualitative shift in the threat landscape: AI is now capable of executing the full ransomware lifecycle at machine speed, making human-centric response timelines dangerously inadequate.
Congress Warned China Is Actively Targeting US AI Labs for Military Espionage
Defense One · Risk
Former US intelligence officials told the House Intelligence Committee that China is deploying a range of espionage techniques to steal proprietary AI technology from American labs, with the aim of accelerating its own military and economic capabilities. The hearing — framed around the 25th anniversary of 9/11 — also highlighted AI’s expanding role in cyberattacks, autonomous drones and foreign influence operations. The testimony underscores that AI laboratories have become critical national security assets, and that the line between commercial AI development and strategic military competition has all but disappeared.
https://www.defenseone.com/threats/2026/07/ai-labs-spies/414929/
EU’s Military AI Exemption Leaves Europe Dangerously Exposed, Economists Warn
CEPR VoxEU · Regulation
A CEPR analysis argues that Europe’s decision to exclude military AI from the EU AI Act has left member states’ defence ministries and intelligence services running on infrastructure they neither control nor can independently audit — a vulnerability exposed when the US ordered Anthropic to suspend access to its most powerful models for all non-US persons in June 2026. The authors warn it could take ten to fifteen years for Europe to develop sufficient sovereign AI capability, and that unlike energy or semiconductors, AI dependency can be switched off globally within hours. The piece calls for urgent interim measures as Washington increasingly acts unilaterally in shaping the rules of AI-driven warfare.
https://cepr.org/voxeu/columns/europes-ungoverned-space-military-ai-and-autonomy-cannot-be-bought