The video below is the full 1.5 hour recording of the UK Parliament Joint Committee on National Security Strategy. The topic of the day for the expert witnesses on Monday 19th October 2026 is Space Resilience. But the evidence after the opening Q&A converges on AI and its risks in coordinated cyber-physical attacks on not just military national security infrastructure, but the same enormous infrastructure dependencies in the wider civil / commercial economy. The latter range from from National Air-Traffic Control (say) to on-line retail operations, large and small. Any one of the vulnerable operations can be an entry point for probes and threats to that whole infrastructure.
The three impressive, articulate witnesses in the order they speak and reinforce each other are:
Will Marshall - CEO of Planet(*)
Jessie Hamill-Stewart - Program Manager of NCC Cyber Security
Julia Balm - Director of Freeman Air & Space Institute
(*) Planet are second only to SpaceX Starlink in rapid deployment and replacement of “proliferated” networks of satellites, resilient in numbers, and in the current national security context are successfully contracted directly to Ukraine (not their dubious US ally).
The whole is recommended and fascinating.
Embedded local copy of full video below with key section starting at ~29:45 but the whole gives the context of the dialogue.
My attention was drawn by a BBC Radio4 Yesterday in Parliament clip of Marshall emphasising that the biggest and most urgent risk to the entire infrastructure - whether by state actors or malicious political-terrorist or commercial-ransom actors (or collusion between the two) - is Coordinated AI Cyber-Physical Attacks.
=====
Meta Commentary: As the (male) CEO of a direct competitor to Elon Musk, Marshall starts in sales-pitch mode, and has already been given speaking priority by the chair over the two (female) expert colleagues on the panel, due to his having some urgent business demands which he must leave early to attend to. However the dynamic evolves very steadily to cooperative teamwork as all three sense the shared need to be honest - without directly criticising or offending the committee or their published strategy - about where the urgent focus really lies. The psychology is as fascinating as the content.
=====
My own experience of AI, beyond theoretical research, has been in algorithmic “smart” systems in built-environment facilities applications since the 1990’s as well as use of current LLM-AI’s as a research tool. My experience with Cyber-Physical Security has been primarily in the nuclear (civil and weapons) sectors. In both contexts there’s a large element of developing and applying (and commitment to) international standards. A slow process.
And following Lee’s comment below, a little more background. My ears pricked-up at the committee evidence, not because it was alerting me to a new consideration, but because the slowing-down / throttling of “Algorithmic Ignorance” has been central to my agenda for the entire 21st century. Whilst our knowledge models are imperfect, and they always are, the last thing we need is agents conceived and controlled by imperfect “techbros” reifying and embedding them everywhere at the speed of light. We need humanity in the understanding loop. We must be very careful what we allow to be automated. It was good to hear the urgency and nature of the risks and the need to address content “sovereignty” aired in a formal UK Gov setting AND with the added psychological angle. (Previously most recently, earlier this week, footnote to this rambling Dear Diary entry for Oct 2026, but #Dysmemics recurring throughout both blogs.)
=====


If You're Wondering What Can Be Done About AI, Start Here https://www.smartsettleresolutions.com/post/what-can-be-done-about-ai-risk