The AIs Have It

The debate on the safety of artificial intelligence development is heating up across the Western world.

Sep 18, 2026

Towering Columns

Writing for The Critic, Andrew Orlowski argues against what he labels a ‘doomsday cult’, instead claiming that artificial intelligence poses relatively little risk to humanity.

Stopping a runaway computer software programme is easy as pulling the plug from the wall. A killer AI would need to be able build and harvest its own power sources and resist a military attack. Such a machine is not feasible at present, or for a very long time.
So why would so much money be devoted to what is an absurd, but admittedly compelling science fiction story?
Khalif believes that existential risk, which the EAs abbreviate to x-risk, is “a regulatory capture mechanism”. The EAs co-opt the language to elbow aside incumbent experts and well established safety practices, and advance their own power base, something she has observed in both nuclear and AI regulation. The gambit “places AI technologists as arbiters of life or death”.
Eschatological narratives like catastrophic climate change and AI going rogue and killing us all both inspire extreme reactions, inducing a kind of mental episode. Sam Altman’s home has been attacked by EA doomers, some EAs went on hunger strike outside Google’s London HQ, and it has even spawned a vegan murder cult. Mostly nonbinary, and describing themselves as “vegan anarchotranshumanists”, the Zizians emerged from rationalist and EA circles, and murdered six people before being arrested.
Both preoccupations fundamentally appeal to narcissism. You alone are blessed with an insight: you alone can see the end of the world. Others can’t, and so the urge to share the prophecy is overwhelming. AI doom has been the more persuasive with our political and media elites, for whom faith in AI is a deeply high status opinion. This is not just because they lack the technical grounding to assess such claims independently, but because they desperately want to believe in the prospect of an overwhelmingly powerful AI. They paint the ghost into the machine.

In The Telegraph, Tom Sharpe argues that recent developments in Yemen put the world at risk, but Britain can positively influence the situation.

As for us, HMS Diamond performed excellently in the Red Sea against the Houthis. Her sister ship HMS Dragon – now in the region – could do the same. With Hormuz in stalemate but still deemed too high a risk for crewed ships to enter, Dragon could usefully head to the Bab-el-Mandeb and work alongside the Aspides warships.
Much will depend on what she is currently doing – she has been working with the French carrier group rather than the Americans – and on whether she needs to protect RFA Lyme Bay, mothership for the UK’s nascent mine-hunting drone force. The last open reporting had Dragon in Mombasa on Aug 11-12, RFA Lyme Bay likewise, so my guess would be they are currently in the Gulf of Oman or the wider Indian Ocean. Dragon could be in the Red Sea in a few days. It’s good to have options – something navies can do much better when at sea and fully worked up.
The real indicator to watch is the same one as in July. If this is about to seriously escalate, one of the two US carriers will start steaming west. Given the stalemate in Hormuz, resuming strikes on Houthi launch positions might well be seen in Washington as an achievable win as the midterms approach.
Perim, then, is a small island of minor tactical importance in a narrow chokepoint. What its capture represents though is a pattern: one of escalation and continued chokepoint challenges – a trend I’d suggest we are only in the early stages of. All the world’s navies need to adjust accordingly, not just rely on the US.

In Conservative Home, Blake Stephenson MP argues for a more detail-oriented approach to reducing legal migration.

That is why I am concerned by Government proposals to replace supervised, in-person English language testing for visa applicants with a remote model. Industry experts have raised significant concerns about the potential for fraud and abuse. If safeguards are weakened, organised criminal networks will inevitably look for ways to exploit the system.
Good immigration policy requires attention to detail. The effectiveness of border control is determined not only by what happens at the frontier but also by the integrity of the processes that operate long before someone arrives in the UK. Visa rules, identity checks, language testing, compliance monitoring, and enforcement all form part of the same ecosystem. Weakness in one area eventually creates pressure elsewhere.
There are no simple solutions to the challenges facing our immigration system. Anyone who claims otherwise is offering false comfort. But there are principles that should guide reform: fairness, transparency, effective enforcement and a clear commitment to the national interest.
As Parliament debates the Immigration and Asylum Bill, we must look beyond the immediate headlines and address the wider weaknesses that continue to undermine public confidence. The objective should not simply be to respond to today’s pressures, but to build an immigration system that is sustainable, trusted and fit for the decades ahead.

Writing for The Critic, James Price argues against Rory Stewart’s position on the rape gangs scandal.

More substantively, the grooming gangs did not stop 10 years ago. It is still happening today. Even Starmer’s administration, who had to be dragged into acquiescing to a national review, trumpeted tens of millions of pounds for extra enforcement, and boasted about thousands of annual arrests.
Similarly, Rory repeatedly mentioned Birmingham, as if this were the only place in Britain where such abuse had happened. The reality is that there have been proven cases of such abuse happening in over 50 towns and cities across the UK. I’m sure readers of The Critic will know all this already, but Stewart clearly doesn’t know or doesn’t care, because the issue is too low status for him to read about.
But even this patrician insouciance for such awful crimes notwithstanding, the scales have fallen from my eyes. Stewart is just not nearly as bright as he thinks he is. I’m sure several years of talking to Alastair Campbell would lead anyone to believe that they were vastly more intelligent than others, but it has turned Stewart into a dolt. For example, he intones at one point that Britain should take just 0.2% of its population every year in asylum seekers. He has chosen this because, I assume, it sounds like a small number to those who still treat him as a serious guy.
But that’s about 138,000 people, over three times the 42,394 actually granted protection last year in our already absurdly abused system. In fact, it’s comfortably more than the 86,000 who applied at all. I wonder if Stewart could open his Afghan rolodex to Home Office so it could start actively recruiting more rural farmers to fill his quota?

Writing for The Telegraph, Liam Halligan makes the case that Andy Burnham’s premiership will be remembered to his ideological loyalty rather than his loyalty to Britain.

One part of the British state which has been on the ball when it comes to the UK’s fiscal weakness is the House of Lords. Two years ago, the Lords economic affairs committee, then chaired by Lord Bridges, published a report on our national debt: It’s Time for Tough Decisions. Last week, Bridges led a Lords debate reiterating the same – which is worth reading in full.
Lord Hill, a Conservative, pointed out that during the 1980s, the governments he advised “developed policies that were coherent”, rather than obsessing over opinion polls and endless focus groups. These days, he says, governments of all stripes yield to “a giant lobbying system” demanding “ever more spending”.
Lords Burns, a former Treasury permanent secretary closely involved with policymaking during the 1976 crisis and for two decades after, remarked that during tough periods, “decisive action was eventually taken, even though it involved difficult and often unpopular decisions”.
Burnham seems incapable of taking tough decisions. He’s a decent communicator and has a talent for political campaigning. But he isn’t the determined, tough-minded leader that Britain so desperately needs.
Future historians may note that his speech at Labour party conference later this month will be 50 years to the day since Britain’s 1976 IMF bailout was launched.

Wonky Thinking

The Institute for Fiscal Studies has released their Green Budget 2026, with one chapter examining the effectiveness of disability benefit, and how to reform it to make savings for the taxpayer.

Key Findings
The share of 16- to 64-year-olds claiming disability benefits increased from 5.5% in August 2019 to 8.2% in August 2025. There have been large increases across all ages and health conditions, with the fastest proportional increases for claims relating to mental health, learning and neurodevelopmental conditions and claims from younger adults.
Rising numbers of people claiming disability benefits has pushed costs up. Spending on working-age disability benefits has risen in real terms from £14 billion in 2019–20 to £25 billion in 2025–26. This, combined with rising spending on other health-related benefits, is the main reason that total benefit spending on working-age adults and children rose from 4.3% of GDP in 2019–20 to 4.7% of GDP in 2025–26 (from £127 billion to £148 billion in today’s prices). This puts total working-age benefit spending as a share of GDP around the level it was at in 2015–16. Working-age disability benefits now make up a much greater share of total benefit spending on working-age adults and children – 21% now, up from 11% in 2015–16. Official forecasts suggest working-age disability benefit spending will rise further to £34 billion in 2030–31 (1.0% of GDP and 26% of non-pensioner benefit spending).
PIP claimants are more likely to report struggling to afford essentials than both disabled people who do not claim PIP and non-disabled people. 51% of PIP claimants are in ‘material deprivation’ (meaning they report being unable to afford five or more basic essentials out of a list of 21 items), higher than the share of people with disabilities who do not claim PIP (33%) and much higher than the same figure for people without disabilities (15%). Those who claim PIP tend to have lower household incomes than those without disabilities, which explains part of the difference, but substantial gaps remain even when comparing households with the same incomes. One reason for this could be that people with disabilities have higher living costs. These gaps in material deprivation provide one rationale for benefits that provide additional income to people with disabilities.
PIP claimants are also more likely to report low life satisfaction than both disabled people who do not claim PIP and non-disabled people. 26% of PIP claimants report low levels of life satisfaction, compared with 16% of disabled people not claiming PIP and 4% of people without disabilities. Again, gaps remain when comparing people with similar levels of income. This could be driven by disabled people having higher living costs, or because of other impacts of having a disability on well-being. Another reason that governments may want to provide additional support to disabled people is to improve their well-being. If this is the government’s priority, it should look to provide the most cost-effective support to improve well-being for disabled people. This could be providing cash in some instances, but it also may be providing in-kind support such as therapy, which has been shown to improve well-being for people with mental health issues.
PIP currently gives the same amount of money to people with very different levels of assessed disability. PIP assessments assign ‘points’ to applicants, based upon the number and severity of difficulties they have with daily living activities and with mobility. To get the highest daily living component of PIP (currently worth £5,960 per year), claimants need to score 12 or more points. 53% of claimants reach this threshold, and 26% of claimants get 16 points or more and 5% get at least 31 points – yet these claimants all receive the same benefit amount. There is a case for tying support more closely to the degree of disability. If done using the current PIP assessment, this would particularly benefit younger claimants and claimants with learning disabilities or cerebral palsy, who are particularly likely to score a very high number of points.
Currently, assessors award PIP based on an assessment of the applicant’s difficulties doing certain tasks. One potential reform would be to also require specific medical diagnoses to qualify. This would introduce an extra, potentially more objective, check on the disability of the claimant. However, requiring any medical diagnosis would only have a modest impact on the number of people who are eligible because 85% of PIP claimants already report having a medical diagnosis and 69% report having multiple diagnoses for different conditions – and these figures would likely rise if PIP receipt required a diagnosis. Alternatively, the government could restrict PIP eligibility to just a subset of medical diagnoses, which would deliver bigger savings, at the cost of more people losing access to PIP. Making PIP conditional on medical diagnoses could remove some of the subjectivity of the current assessment, although medical diagnoses also often require some subjective judgement. It would also put additional costs on both applicants and the health system and may push more people to seek private diagnoses for health conditions.
Another commonly discussed option is for PIP to treat mental and physical health conditions differently, perhaps substantially reducing entitlements for the former. Administrative data show that 45% of existing PIP claimants have, when assessed, a mental health condition or a learning or neurodevelopmental condition listed as their main disabling condition. One way of restricting access to PIP for those with mental health conditions would be to require that claimants have a physical health diagnosis. This would significantly reduce the cost of PIP. However, these savings are both uncertain and likely far smaller than that 45% headline might suggest, for two reasons. First, two-thirds of claimants with a diagnosis for a mental health, learning or neurodevelopmental condition (irrespective of whether it is the ‘main’ condition) also report a physical health diagnosis. Second, more claimants would likely obtain a physical diagnosis if it was a requirement for PIP receipt. If the government wanted to make bigger savings, it could restrict access only to those with certain physical health diagnoses.
The government could reduce the generosity of the disability benefit system specifically for young people, perhaps if it considers any negative effects of receiving these benefits to be larger for those who have a whole working life ahead of them. There are 689,000 people under 30 who claim PIP (20% of the total working-age PIP caseload), accounting for £5.5 billion per year in spending. However, young PIP recipients are disproportionately likely to have the most severe disabilities; 12% of 16- to 19-year-old recipients receive at least 31 points in the PIP daily living assessment, compared with 5% among all PIP recipients. If policymakers wanted to keep providing PIP for young people with more severe disabilities (proxied as those who get the highest possible award) while removing it for others, the impact on the caseload and spending would fall to 327,000 and £2.2 billion respectively, before accounting for any behavioural responses.
One unusual feature of PIP relative to other benefits is that it is not means-tested, so high-income households can claim it. If the main goal of the benefit is to reduce inequalities arising from disability, there is a good case for keeping PIP as a non-means-tested benefit. Alternatively, if greater weight is placed on ensuring that disabled people can obtain a minimum standard of living, there is a stronger argument for means-testing PIP – something recently proposed by Reform UK, though the Timms Review has indicated it will recommend keeping PIP non-means-tested. Means-testing PIP, by integrating it into universal credit, would deliver a saving of £8.2 billion (33% of PIP spending) if claimant behaviour did not respond to the change. However, in reality, the saving will be somewhat smaller due to behavioural responses. In response to the change, more people would claim universal credit in order to keep their PIP, and fewer people would work as the reform would weaken work incentives.

Podcast of the Week

Outpost Studios have released a podcast with Colin Brazier examining the legacy of the September 11th attacks, and how the West is still largely unwilling to confront the ideology behind the attacks.