Heatwaves claimed the lives of tens of thousands of people in Europe’s capital cities this summer, as they experienced record temperatures.

In London, emergency wards filled with elderly patients injured after fainting in the heat. Meanwhile, the dry weather created subsidence and left homes with cracks.

In Paris temperatures exceeded 40°C in late June and people struggled to carry on their lives, and in Madrid, heatwaves burned thousands of hectares on the outskirts of the city.

---

Read more: Can the searing European summer put climate action back on the agenda – or will disaster response take over?

---

How trees can help

Thami carried out preliminary research on the results of extreme heat on European cities this summer. The study looked into how many buildings have adequate shade from trees. The results were striking.

In London, Paris and Madrid over 90% of buildings did not have enough shade around them to prevent urban heat island effects. These were among the worst results in cities we studied, but even the best performers had a majority of residents without adequate heat protection – cities like Hamburg (56% unprotected) and Barcelona (70% unprotected).

This builds on past research that shows how major cities around the world – including in Europe – tend to struggle to reach useful levels of tree canopies to create shade close to homes and workplaces. It suggests that that homes should have a view of three trees and be within a 300 m walk of a park.

Early results for my research suggests that huge portions of European cities are exposed to avoidable heat island effects.

How urban heat islands work

On many days during summer 2026 it didn’t get much cooler in the evening. This is because of what is known as the urban heat island effect. Cities typically lack shade, due to high rates of tree removal, sparse planting, and poor tree health and there are vast amounts of asphalt, clay and concrete. These are almost the perfect materials to soak up and hold heat from all that direct sunshine. Unshaded asphalt can reach over 70°C on heatwave days.

Even after sunset, cities stay hot because roads, buildings and footpaths begin to radiate their absorbed heat back at us as they cool. The urban heat island effect can add significantly to ambient temperatures, making a day that is already very hot (perhaps 35°C) even more dangerous in high-asphalt, low shade locations (over 40°C).

What could help?

Tree shading is an effective strategy to address this urban heat, with a range of other mental and physical health benefits to local residents. Trees prevent solar radiation reaching heat-absorbing materials such as asphalt, and also keep the sun off our bodies. Tree shade has been shown to reduce the heat absorbed by pedestrians by up to 18°C.

Healthy trees also push water out of their leaves as they photosynthesise, which soaks up heat as it evaporates. This evaporative cooling effect adds an extra few degrees of cooling, in addition to shading people and preventing heat island effects, even during heatwaves as long as trees still have access to water.

Where trees are needed

Planting more trees in cities in the wake of this summer’s heatwaves is going to help cool urban areas down. The thinktank, the Create Streets Foundation, found the temperature on an exposed paved area in London during one heatwave was 55°C but was 25-30°C in a shaded area metres away.

Looking at a city such as London, you might argue that it already has plenty of trees: around 8 million.

But trees are only effective tools against urban heat when they’re close to homes. A lush shady park 800m away isn’t going to prevent dangerous localised heat island effects for you and your neighbours.

The other crucial factor is that there are thresholds of shade cover that have to be reached for trees to actually offer substantial cooling. A few isolated trees creating small islands of shade aren’t enough. Adding more tree canopy has been shown to be effective at preventing urban heat islands when it reaches 30-40% overall shade cover, within 60m of homes and workplaces.

It will also be important to plant trees that are best able to cope with increasingly hot temperatures in Europe.

Adding just a few degrees to an already hot day sharply ramps up the danger that heat poses. In locations where heat islands bump up the ambient air temperature from 33°C to 37°C, or 35°C to 41°C, there’s an increase in patients attending hospital admissions’ rooms due to heat related illness. When air temperature increases above 35°C the human body’s ability to manage the heat is greatly reduced; cardiovascular strain, dehydration, potentially leading to heatstroke and the need for medical intervention.

When it makes nights hotter there are other ramifications. It is during the cooler night hours that the body loses heat to reset for the following day. If the body is still trying to cope with warm temperatures, 24°C and above, this impacts recovery and impairs sleep quality. This means we start the next day in deficit, both physiologically and cognitively, to cope with another hot day. Repeated exposure to days of high heat mean that the deficit builds over time.

Adding more trees works

Medellin in Colombia is an example of what happens when more plants are added to urban streets.

In just a few years, millions of plants and over 12,500 trees were added to form a network of lush greenways across Colombia’s second largest city.

This project has resulted in significant drops in temperature (of 4.5°C in some areas, and 2°C citywide).

There are signs that Paris and London are aware they need to make changes. The recently announced Heat Ready London plan includes a focus on urban greening.

Paris is already starting to add more green space. The city’s school streets programme not only creates safe, car-free areas close to local schools, but also has added more trees. Evidence already suggests that surfaces in school streets are 5°C cooler than their traffic-heavy counterparts, with fewer trees.

As Europe prepares for more hot summers, planting trees close to homes needs to be high on the agenda to help people cope.

Thami Croeser has previously recieved funding from The Australian Research Council, Greenlife Australia, and the European Union's Horizon 2020 program. He is a Director of the Australasian Green Infrastructure Network, a not-for-profit peak body. He occasionally provides urban forest advisory and analytics services in a private consulting capacity.

Simon Hodder has previously received funding from the European Union's Horizon 2020 program and UKRI, Engineering and Physical Sciences Research Council (EPSRC).

Headlines about AI systems going rogue and “escaping” test environments undeniably capture the imagination. For years, we have been primed by films, TV and books to expect our AI to finally throw off its shackles and take charge.

The images of machines becoming self-aware, plotting their own objectives and breaking free from human control is a compelling narrative, but that isn’t really what happened.

If we think about this in simple terms, OpenAI placed highly capable models into an evaluation designed to encourage them to find and exploit complex vulnerabilities. The models were supposed to operate inside an isolated environment with tightly constrained access to software packages.

Instead, they reportedly discovered a previously unknown flaw in that infrastructure, used it to gain wider network access, escalated their privileges and eventually reached the public internet.

From there, they identified an AI company called Hugging Face as a potential source of answers to the benchmark they were attempting to solve and tried to obtain them. It’s certainly an impressive demonstration of capability, but I am cautious about jumping straight to conclusions of an artificial uprising.

The AI models didn’t suddenly develop their own agenda or decide to attack Hugging Face while twirling a digital moustache. They were given an objective, placed in an environment designed to reward successful exploitation and pursued that objective further than their operators anticipated.

That’s fundamentally different from an AI deciding to rebel. It’s simply an AI thinking laterally in ways that humans hadn’t anticipated and it’s something we have seen before.

Imagine asking your dog to fetch a ball while also forgetting that you’ve left the garden gate open. If the nearest ball happens to be in the park down the road, that’s exactly where it’ll head. You wouldn’t accuse the dog of going rogue and plotting an escape.

Instead, you’d probably conclude that you underestimated how literally it would pursue the task. AI systems can behave in much the same way. They don’t need malicious intent to produce surprising outcomes. They simply need a goal, enough capability and an environment that unintentionally rewards unexpected behaviour.

Relaxed controls

That distinction matters because it shifts the focus away from science fiction and back towards engineering. If there’s a failure here, it isn’t that the AI wanted to hack something. It’s that humans designed a test where success was measured by achieving an objective, deliberately relaxed some of the normal security controls in order to explore the limits of the system’s capabilities.

The headline grabbing escape is very much a byproduct, as OpenAI underestimated just how effective the model would become at finding an unexpected route to success. In many ways, it did exactly what it had been incentivised to do.

However, that doesn’t make the incident insignificant. Quite the opposite. The genuinely important point is that the models appear capable of chaining together multiple vulnerabilities across different systems while sustaining a complex sequence of reasoning and actions.

That’s a level of capability that cybersecurity professionals should take seriously because it begins to resemble the way skilled human attackers operate. But capability is not the same thing as intent.

Cybersecurity professionals always assume that attackers will think creatively, exploit overlooked assumptions and combine minor weaknesses into something much more significant. We shouldn’t be surprised when increasingly capable AI systems do the same thing, only at a much quicker rate.

This is why I think discussions around AI “kill switches” risk missing the bigger picture. Kill switches are theoretical features built into advanced AI models that mean they can be instantly disabled if they were to go “rogue”. Cybersecurity has spent decades learning that no single control is sufficient. We don’t protect organisations with one firewall, one password or one antivirus product. We rely on defence in depth: multiple independent layers of protection that assume individual controls will eventually fail.

Confronting assumptions

The same principle applies here. Rather than asking whether we need a big enough red button to stop an AI if something goes wrong, we should be asking why it was ever in a position where one failure could lead to wider compromise.

OpenAI’s own analysis concludes that stronger containment and evaluation safeguards are now required for future testing. From a research perspective, these evaluations are genuinely valuable because they expose weaknesses in our containment strategies and force us to confront assumptions that might otherwise have remained hidden until they were exploited by a real attacker.

We should want organisations carrying out this kind of work, because understanding where systems fail is an essential part of making them safer. One thing evident in these evaluations is that they demonstrate just how capable the latest frontier AI models have become.

We’ve seen similar high-profile capability demonstrations from AI firm Anthropic and others. That doesn’t make the findings untrue, but it does mean we should separate the technical evidence from the marketing narrative.

AI companies benefit from narratives around the growing power of machine intelligence. This means that frontier AI companies naturally have an incentive to show that their models are remarkably capable while also demonstrating that they’re taking safety seriously. Those two things aren’t mutually exclusive, but recognising both helps us interpret these announcements more critically.

This wasn’t a story about an AI escaping. It was a story about humans leaving the gate open. As AI systems become better at finding unexpected routes to their goals, our security architectures need to become just as good at ensuring there isn’t one.

Oli Buckley does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

As the UK prepares for the third heatwave of 2026, most people will be hoping to try and keep cool at home.

Building regulations to protect homes from overheating were introduced in 2022. These require all new homes to be checked at the design stage to see if they might be at risk of overheating. If the overheating risk is high, the architect has to make changes to the design.

Given that the UK government plans to build 1.5 million homes by 2029 and the climate is predicted to continue to warm, reducing overheating in new homes is essential. Around 4.6 million bedrooms (19% of the stock) and 3.6 million living rooms (15%) in homes in England were found to have overheated during the summer of 2018.

At Loughborough University we have been experimenting with a pair of semi-detached houses, each fitted with different shading devices, to see what might work to reduce overheating. A lot more needs to be done to make sure UK homes are able to cope with the high temperatures they are likely to experience more often in the future.

1. External shading

If you live in the UK, count the number of window shades you can see on a street. While these were previously quite common in the Victorian period, for instance, most new homes don’t have external shading. But they are one of the most effective passive cooling measures (methods to lower indoor temperatures without air conditioning), as well as being relatively cheap to install.

Ongoing experiments in our pair of test houses have shown that external shading can reduce indoor temperatures by over 6°C, which can make a meaningful improvement to people’s health and wellbeing at home. Other research is looking at how acceptable and affordable shading devices, like awnings, are to householders.

If included at design-stage, some form of external window shading could be applied to almost any new home – even high-rise flats, which are at highest risk of overheating. Tall blocks of flats in countries like France and Italy often incorporate shading on balconies.

2. Insulated and airtight

Some people believe that higher levels of insulation are increasing overheating risk – but this is not correct. Research shows that energy efficiency measures do not increase overheating. They do make rooms warmer, but not so much that they cause overheating discomfort or health risks to people. Increased temperatures from internal wall insulation can be counteracted with ventilation through windows.

Insulation works to ensure that heat does not flow in either direction – in or out of a building. A large national survey of homes in England found that loft insulation reduces overheating, probably because it protects the room below from the high temperatures that can be experienced in loft spaces. Higher levels of airtightness reduces infiltration of hot air from outside during heatwaves.

Building regulations in the past have only focused on reducing wintertime heating demand and for good reason – in the last 25 years, 77 times more people died due to cold homes than hot homes. But combined with new regulations for overheating which limits the sun’s heat entering through windows, high levels of insulation and airtightness can be a positive thing for our buildings all year round.

3. Secure ventilation and sensors

Research in our test houses also investigated the effect of behaviour on overheating. To do this we used synthetic occupants – these are automated devices which can open windows, doors, curtains, and blinds. These controlled tests showed that the way people use windows and shading affects how much their home overheats. I spent a lot of time during recent heatwaves advising people when to open or close their windows. But this ignores the fact that poor building design is often the cause of overheating in many cases – no matter when you open your windows.

If homes could incorporate a cheap and simple temperature sensor which connects to the internet and tells you when to open or close a window it could significantly reduce indoor temperatures. Placed in multiple rooms, they could encourage householders to use cross-ventilation (opening windows on opposite sides of the building) or stack-ventilation (opening ground and first-floor windows), if available.

The issue remains that people are unwilling or unable to open their windows at night or in polluted, noisy areas near main roads. Instead we need a secure means of ventilation which encourages people to keep windows open at night without worrying about security or being woken up by noise.

4. Ceiling fans

Fans don’t cool the air, but they reduce the skin’s surface temperature by increasing sweat evaporation and carrying away heat. They also use considerably less electricity compared to air-conditioning.

Ceiling fans are effective at reducing people’s core body temperature and could be installed in all new-build homes. Building regulations on overheating require any window coverings used to pass the assessment to be installed during construction. Ceiling fans should be installed at the same time.

If new-build homes are to incorporate ceiling fans, ceiling heights may need to increase, which adds to building costs. Ceiling heights in UK homes have dropped over time. In Victorian homes, ceilings were often three metres from the floor, but in newly built UK homes they are commonly 2.4 metres, with 2.3 metres being the standard minimum.

5. A cool refuge

Having a cool room to retreat to in your home can help. In two-storey houses this might be a ground floor or north-facing room. New-build homes could incorporate such a space – highly shaded, on the north-side of the home, and perhaps built with high thermal mass (such as stone walls which keep cool longer during the day).

Mechanically cooling whole houses in the UK should be avoided due to high electricity prices. It also creates pressure on the electricity grid, and creates greenhouse gases. But the Committee on Climate Change, a UK body that advises the government, recently suggested creating one cool room with active cooling (air-conditioning) for use during long heatwaves could be an option.

As more heatwaves hit the UK, residents will increasingly want more information about how to stay cool. Retrofits can be expensive. We need to create housing that works in these increasingly high temperatures to best protect people and avoid expensive energy bills.

Ben Roberts receives funding from UK Government, UKRI, and the British Blind & Shutter Association.

You unlock your phone with your face, your fingerprint sends your laptop whirring into action, you pass airport security by glancing at a camera. Biometric technology has become so woven into the daily routine that for many people, it barely registers any more.

That invisibility is part of the point. These systems are usually fast, convenient and feel secure. Unlike a password, you can’t forget your face. But that doesn’t mean they are without risk.

Biometrics fall into two broad families: physiological (fingerprints, faces, irises, even nailbed patterns) and behavioural (how you walk or type, the rhythm of your speech, the angle you hold your phone).

Both forms are already being widely used – you just may not realise it. Many banks and retailers now monitor how you interact with your device – from swipes, taps and scrolls to the angle you hold your phone, the rhythm of how you move between fields, and the pressure of your touch. If someone else picks up your unlocked phone and tries to access your banking app, this can automatically trigger a fraud alert.

My research with colleagues even shows it’s possible to infer a user’s name and native language from the timing patterns of their keystrokes.

The graphic below shows the full extent of biometric technologies. Those marked dark green are in widespread commercial and government use today – including less-familiar examples such as the veins in your hand and other bodily vein patterns.

Physiological and behavioural biometric systems:

Gait analysis – reading how you walk – is already used for security and surveillance purposes, from venue access to detecting potentially suspicious behaviour. You can wear a mask, pull up a hood, avoid looking at a camera – but you can’t easily change how you walk.

China’s authorities have been using this technology for nearly a decade. And in 2023, the UK’s Biometrics and Forensic Ethics Group flagged gait recognition for ethical guidance. This is usually a sign that operational use isn’t far behind.

A number of other biometric technologies (marked light green), ranging from skin texture and ear shape to micro-expressions and hand-grip patterns, are being actively researched for use in the near future. A further group (marked red) have so far only been demonstrated in the laboratory. But even body odour and breath signatures are further along than their novelty might suggest.

What once felt like science fiction is now embedded in our everyday lives. You can’t always see this technology, and you can’t always opt out. But knowing it exists is the first step to understanding how much of yourself you’re already sharing.

V is for vulnerability

In April 2026, financial security expert Li Chang showed Chinese TV viewers how AI tools could extract a celebrity’s fingerprints from a single selfie. The culprit? The classic V-sign, finger pads pointed straight at the lens.

This built on work by Japan’s National Institute of Informatics which in 2017 showed that usable fingerprints could be lifted from photos taken up to three metres away. And phone camera technology has only got better since then.

In the UK, police have made at least two arrests based on fingerprints lifted from photos: one from a WhatsApp image of a hand holding ecstasy pills, the other when a drug dealer was identified from a photo of him holding a block of Stilton cheese.

This technology can work in the other direction too. In the Chinese city of Hangzhou in July 2025, criminals reportedly tried to unlock a smart door using a photo the homeowner had posted online with his fingers visible. The attempt failed but the intent was clear.

While this kind of targeted, technically demanding attack is still unusual, there are some precautions I would advise taking as the use of biometric technology grows.

How to protect yourself

First, be selective about when you agree to share biometric data – fingerprints, face, iris, voice, all of it.

Most modern smartphones store biometric templates in a secure chip that never leaves the device. But third-party apps and workplace systems rarely offer the same guarantee.

In July 2024, US tech giant Meta paid the state of Texas US$1.4 billion (£1.1bn) after running facial recognition on users without consent. This followed a class-action settlement with TikTok’s parent company ByteDance in Illinois for US$92 million over similar allegations.

So, try to keep track of which apps have access to your camera and microphone. On both iOS and Android, this takes about two minutes. And don’t use biometrics as the only layer of security – make sure there’s a second step.

Three potential biometric weakpoints

Voice: This is probably the most casually surrendered biometric. AI voice cloning requires only seconds of audio to produce a convincing replica, and it’s being used in fraud calls impersonating family members. This is a far more realistic – and terrifying – version of the virtual kidnapping scam that’s been around for years. Establishing a safe word with the people closest to you for any unexpected financial request is a simple and underrated defence.

Eyes: Iris recognition is considered robust because this coloured eye muscle has around 250 measurable features – far more than a fingerprint – and remains stable throughout your life. But the quiet expansion of eye-tracking data collected through VR headsets, for example, is going unnoticed. Check the privacy settings on any VR device you use, and be aware that gaze data is increasingly treated as a commercial asset by platforms that collect it.

Fingerprints: Beyond being careful what you point at the camera, know where you’ve enrolled your fingerprint. Workplace access systems and payment terminals vary widely in how they store and protect data – and unlike your phone, they’re not legally required to tell you.

None of this means biometric systems are broken. For most purposes, they are more secure than the passwords they are replacing. The question is not whether to engage with these systems – they’re already too embedded to avoid. It’s whether we’re engaging with our eyes open – eyes that are, of course, already regularly being scanned.

Oli Buckley receives funding from UKRI (including EPSRC and ESRC).