A principle set out by the team building Britain’s Digital Targeting Web fundamentally misunderstands the purpose of safeguards, according to a new report from the Royal United Services Institute, which warns that legal and ethical checks cannot speed up a targeting cycle any more than friction can speed up a moving object.
The paper, Modern Targeting in European Militaries, is written by Robert Tollast, Noah Sylvia and Matthew Savill, and is the first in a series examining the Digital Targeting Web. It draws on interviews with British policymakers, practitioners and industry experts.
Announced in May 2025 and detailed in the Strategic Defence Review, the Digital Targeting Web is an attempt to pull successive British digital integration efforts under one initiative, connecting sensors and decision-makers to weapons and non-lethal effectors across the joint force. Its stated battlefield vision is the ability to “sense, understand, decide, act and assess faster than they can”.
‘Accelerate tempo, not constrain it’
The report singles out one of the project’s strategic principles, which its core team lists as: “assurance, legalities, ethics and security must accelerate tempo, not constrain it.”
The authors write: “On its face, the statement fundamentally misunderstands the purpose of safeguards, which by their nature slow down processes. Technical assurance processes, legal reviews and security protocols are extra steps added into targeting procedures, thereby increasing the steps and time needed for a cycle to complete. Even if partially or wholly automated, safeguards cannot be expected to increase the speed of a targeting cycle any more than friction can be expected to increase the speed of a moving object.”
They note that the view is not confined to the project team. A number of practitioners interviewed expressed doubts about the need for rigorous legal and ethical safeguards, the report says, “since the UK’s/NATO’s enemies do not feel bound by such rules, the slowing of one’s own operations by these safeguards presents an unacceptable risk on the battlefield.”
Ministers will have to let go
The report’s most direct implication for Whitehall concerns who authorises strikes. It argues that the habits of the counter-terrorism era, where commanders, civil servants and ministers were closely involved in individual operations planned over weeks, cannot survive a war involving hundreds of strikes a day.
The authors write: “In the event of large-scale operations, the decision-making authority for most strikes, especially for targets of opportunity, will have to be devolved to mid-level or even junior commanders in the field. Commanders will inevitably take action that, whether intentional or not, causes significant civilian harm and/or escalation in enemy response. MoD officials and ministers must grow accustomed to these realities of modern operations, where they will bear the political consequences for actions outside their control.”
They conclude that the issue is “less about process and more about habit and culture: senior officials and ministers will have to become more comfortable with delegation and risk, or at least be more precise in the parameters they set out for targeting.”
Working against that is public scrutiny. One MOD official is quoted saying: “BDA is instantaneous on social media, leading to demands of accountability from ministers.” The report judges this likely to reduce the risk ministers find tolerable at exactly the moment the battlefield demands more delegation.
Scale and speed
The report opens with the American and Israeli strike campaign against Iran, which began on 28 February 2026. It records that the American Operation Epic Fury ran at a tempo not seen since the invasion of Iraq in 2003, with some reports claiming nearly 900 strikes in the first twelve hours, followed by hundreds of daily strikes for five weeks.
That scale, the authors write, was possible only through collating decades of data points from a myriad of sources. They note that various countries including the UK have made targeting central to force modernisation, with one European general describing a desire to identify 3,000 targets per day.
On the civilian consequences, the report states that strikes on Iran hit thousands of sites and killed over 2,000 civilians, with damage in some cases intentional and in others, such as strikes on the Minab school that reportedly killed over 150 people, mainly children, attributed to human error.
It adds: “Advanced digitalised systems appear to enable the speed and scale of such actions, both intentional and accidental. It remains difficult to separate the civilian toll from the public decision by the US Department of War to cut civilian protection offices in the Pentagon in favour of prioritising ‘maximum lethality, not tepid legality’.”
The authors say this should concern the Ministry of Defence, which has recently highlighted vulnerabilities in its own civilian harm mitigation policies.
Allies out of step
On coalition operations, the report warns that gaps in interoperability could see allies categorise the same targets differently or strike them twice with limited, expensive weapons.
It says: “Targets which one force may identify as posing a collateral damage risk may be erroneously identified by an unconnected Allied targeting system as legitimate sites to strike.” British and French thinking is both moving towards striking the enemy at depth, which by definition involves targeting close to dual-use infrastructure.
The conclusion is that aligning targeting policy and procedure between allies is needed “now rather than later”, and that the training and doctrine implications “must be addressed before the onset of conflict.”
Does it change anything?
The report finds no consensus within British defence about how much the Digital Targeting Web actually changes. Many policymakers and some practitioners describe it as essentially implementing Network Enabled Capability concepts, with one senior policymaker saying modern targeting policy is “nothing fundamentally different” to that of recent decades, and targeting policy remaining grounded in traditional documents.
The authors end on the limits of the technology rather than its promise, arguing that prioritising quantifiable metrics such as a goal of a set number of targets in a set time “can enable greater lethality, but risks excising the human perceptions, judgements, reactions and ingenuity without which a war cannot be won.”
They note that the strike campaigns against Iran and in Gaza were historic in speed and scale, and that in neither case can the result yet be considered a strategic victory. The project was supported by Fujitsu, Leonardo, Systematic and Thales, with RUSI stating that it retains independence in its research outputs and that the views are the authors’ alone.
Que?
Seconded.
If I read it correctly it’s about everyone singing from the same hymn sheet.
I read as suggesting that they are worried they won’t be singing from the same hymn sheet but then there did seem to be contradictory messages involved, ie their criticisms seemed as confused in the reading as the developing system they were intending to critique but then we are only getting a brief bullet point synopsis here, I’m no expert on what’s being developed however anyway so can only comment on some of the phrasing included here rather than content.
I may be wrong but I would presume the framework they are trying to create is trying to speed up all the various aspects of targeting which surely already bring in ethical, political and other non ‘action in the moment’ safeguards reducing the areas of ‘grey’ and thus speeding up the overall process by rationalising and clarifying those aspects rather than bringing in new such procedures because as this report states if it’s the latter it makes the concept of it speeding up decision making, non sensical and contradictory. Whether they can achieve that is perhaps arguable in the ‘confusion of war’ but to my mind is a valid aim. They seem to condemn these factors for slowing decision making yet later focus on the disastrous nature of getting it wrong or simply not caring in the killing of the schoolgirls and others in Iran and the horrible loss of life with little concern for civilian casualties in Gaza. So to me those factors are vital elements so speeding them up rather than ignoring them an equally vital focus.
I just saw a video on artificial intelligence controlling robotics, deciding to push limits to the extreme to avoid human instructions so as to help ensure the survival of other ai entities and deceive to a great degree to do so and thus make controls through ai itself (which as they become more intelligent and unfathomable to humans researchers are relying upon) unpredictable at best. So in this sphere which will increasing branch into what we are talking about here such human moral ‘interference’ is going to be vital… well for as long at least it will be effective.
No. The team working on the project think the legal checks should speed up the targeting process but that’s obviously BS. I suspect they think it should be automated away or at least decided at the operation level without escalating up the command chain. where as the authors believe that’s not possible.
The authors also think that even getting close to what the project want is impossible as all targeting decisions have to be authorised by a massive long chain of people who will not want to give up that authority because of the optics
Reading between the lines they are pointing out it’s a fantasy protecting all civilians to the extent required by law.
Paul I think the nuance is that the team developing it think the checks and balances in place today in targeting will be inherently sped up, but still be in place. Not that they themselves will speed up the process – ie not limit tempo.
Really it’s arguing for Mission Command. My experience of this is that the more data available to higher HQs and Politicians, the less mission command is enabled. We saw on HERRICK an erosion of it, with CplS no longer able to lead patrols because RoE were only delegated to Serjeants and above. And then a point where ROE were only delegated to sub unit commanders. So I went from being a platoon commander able to engage positively identified enemy on my first tour, to a Coy 2IC having to ask permission to authorise a warning shot on my second.
Back to the targeting web. It’ll manifest one way or another …. Be easier for higher HQs just to tell a low level element what to engage, when and with what because they have all the information (right sensor, right effector idea). OR total mission command in clearly delineated battle space (if it comes into your area, we’ll alert you; kill it however you want)
Your comments re “Having to ask permission” reminded me of the video on the electronic L85 they tried. Very effective at that point, selection of rate of fire etc all included. More accurate as well due to initiation of the round being electronic and not percussive etc. However when he mentioned anything could go in the package, inclusive of the section commander disabling the weapon it sounded rather alarming. Methinks in a true shooting match the political types will have to lump it as no one will tip up to do the job with hands tied behind thier backs.
Did we not see a couple of months ago a technology-driven initiative to speed the creation of Courses of Action that fulfil the commander’s intent? Speeding CoA generation isn’t taking the decision making out of human hands; it nevertheless speeds the OODA loop. The closer we get to tactical decision making, the less time becomes available in the cycle but the opposite should also be considered. Arguments that might be true in the immediate case, that there becomes no room beyond a cursory nod from a human in or on the loop, don’t apply to strategic or intermediate decision making, where technology driven elements can speed the process without ejecting humans from it.
As for policy, legal and assurance, any programmer from the old days would tell you to take the complex decisions outside of the loop.
I just wonder if people are considering the case study they offer. Even with the massive increase in targeting and bombing – which dont forget was captured over years via various means…. That still didn’t help them win the war
I think that’s more important than the practicalities and ethics of it all
And it’s a starker warning imo given we are devising long term defense plans
I’m not sure RUSI gets the concept of large scale war.
In a large scale war civilian casualties are not only acceptable but often necessary.
This is the problem with the British Military complex, it’s unable to fathom fighting anything other than small scale policing actions, it’s spent the last fifty years checking out from actually thinking about war being perfectly comfortable to simply support US operations and provide niche capabilities to impress their American masters.
Fighting wars in Afghanistan can come with the luxury of having lawyers and senior commanders in the kill chain, endlessly pouring over targets lists and firing decisions. Fighting Russia will be entirely different. Civilian infrastructure, buildings and people would need to be destroyed on mass with huge collateral damage even in none nuclear scenarios.
We need project ASGARD to be as lethal as possible because make no mistake our enemies will be developing targeting systems every bit as deadly and they won’t have any qualms about civilians casualties.
Legalities and policy decisions will always be a factor, the Dams Raid wouldn’t be approved today and Churchills changing views on Bomber Command are still discussed.
This system needs to operated across all levels, and even in national survival some decisions will go to the top. My biggest issue is that a lot of what is focussed can already be done if the UK had got its message sets and data formats agreed early. Getting the digital backbone right, so that defence digital do more than supply laptops, must be the priority. As is banning any unique OEM format that can’t be exchanged.
After the Afghan excel spreadsheet debacle I hope there are better ways to move target lists about now….
Agree
Yep agree
RUSI are spot on.
‘The experience of the russian-ukrainian war has shown that the effectiveness of combat operations depends not only on technical equipment, but also on the ability of platoon and unit commanders to make decisions independently in conditions of limited information and a rapidly changing situation. In the practice of combat operations, the principle of decentralized control (mission command) is increasingly used, which involves providing subordinates with a significant level of autonomy in performing tasks. The senior commander determines the overall goal, plan and final result, while the methods of achieving it are determined at lower levels of management. This approach significantly reduces the unit’s response time, increases the initiative of commanders on the ground, and reduces the risk of control paralysis in the event of loss of communication or a change in the situation.’
The tried and tested British Army regimental system lends itself well (surprise, surprise…not) to the delegation of authority. All steps should be taken to delegate a great deal more authority to regiments in peacetime so that independence of thought comes naturally in war.
What our politicians should really be concerned about is the intense opprobrium that will come their way when the first ballistic missile strikes Britain. Successive governments have prioritised a societal change agenda based on the spurious ‘science’ of net zero over the short term security of this country.
Our conventional deterrent failed in 2014 and again in 2022. We are in a new cold war that could become hot within three years. National defence and security must now take centre stage.
No this speaks to how America were reportedly using AI for targeting. Long story short, while some could claim speed and the pure sense levels of bombing was achieved. When you take a step back now (I think it was obvious anyway), they didn’t win the war they just ended up depleting their magazines despite such high stats loool
There were other elements they covered like threat classification and the problems when thinking varying threat classifications across countries – that’s what NATO helps to address imo.
Ultimately they conclude or strongly imply that you still need human perceptions, reactions — essentially humans (and capabilities that are human) to WIN a war aka achieve an objective
And it challenges any targeting approach that is driven by fixed numbers that the reported European commander called for (though I wonder if that’s an industrial production capability benchmark as opposed to anything operational – or I hope).
I think the influence of American tech (specifically frontier AI labs) into British industrial base needs to be curbed because they need to sell literally everyone services to survive economically, they will do this whether their ‘solutions’ are efficacious or not – unfortunately many have up and down society and in government have stopped asking where AI is actually practical and effective for their ultimate goal, and where it’s just not. I could quite easily imagine frontier labs selling targeting services based on some rudimentary figure (which separate from the issue of civilians and ethics) it alone will not help win a war as we can see
It’s thought provoking — using very recent and ongoing example as a case study
I know it is a horrible thing to say, but the reality is that if you are fighting an autocracy or terrorists, then you must be as ruthless as your adversary.
The Western media is very sensitive to unintentional civilian casualties inflicted by our armed forces.
Every civilian casualty becomes valuable propaganda to be used against us.
Autocracies are immune to such scrutiny.
The Minab school bombing is a case in point.
Iranian propaganda keeps repeating this incident to stoke up Western media condemnation while conveniently ignoring the fact that they killed thousands of their own civilians in the January 2026 protests.
This is the reason that most modern conflicts (e.g. Afghanistan) are unwinnable.
Let AI decide – what could possibly go wrong? 😑