Your phone suggests replies before you've finished typing. Your watch tracks your "biocharge." The speaker in the kitchen is waiting for a question, and the Wi-Fi router hums away while cloud-based AI runs machine learning models somewhere far beyond your front door.
AI is now a common part of everyday services and workplace systems, rather than a specialist tool used by a small number of tech firms. Stanford’s 2026 AI Index reports that 88% of surveyed organisations use AI in at least one business function, while 70% report using generative AI in at least one part of their business. As AI becomes more common in search, communication, customer service, shopping and smart devices, it also increases demand for the connected hardware, networks and power infrastructure that support it.
So, does the growth of artificial intelligence mean higher EMF levels where you live and work? The AI EMF question is whether this growth means more exposure at home and work. The honest answer isn't a clean yes or no. AI software doesn't create an electromagnetic field independently. The relevant sources are transmitting devices, electrical infrastructure and AI data centers, rather than the software itself.
Cloud AI may run in a data center, while ordinary home devices connect through wireless technologies, including Wi-Fi, Bluetooth and mobile networks. These signals are forms of non-ionizing radiation. AI may increase connected-device use and electricity demand, but that doesn't prove a universal increase in personal exposure.
Key Takeaways
-
AI is software, not an EMF source itself. The devices, networks and electrical systems that support it are responsible for producing RF and EMF.
-
More AI may mean more connected devices, data traffic and power demand. It does not automatically mean higher EMF exposure everywhere.
-
Large data centres can add local low-frequency sources through transformers, switchgear, UPS systems, bus ducts, feeders and substations.
-
Fields are usually strongest near electrical equipment and power routes. Conditions beyond a site depend on its design, load, distance and shielding.
-
Measure specific sources where useful. Use distance and wired options to reduce unnecessary exposure, and assess each site individually rather than making assumptions about its wider neighbourhood.
Does the Growth of Artificial Intelligence (AI) Mean Higher EMF Levels?
AI is a computing process. It needs processors, servers, networks and high levels of electricity, but machine learning models have no invisible force field of their own. AI growth may contribute to higher EMF levels where more devices transmit, more network infrastructure is installed, or electrical systems work harder.
That distinction matters. A phone running an AI assistant, a smart camera uploading footage and a data centre training a language model are different exposure scenarios. Distance, power output, frequency, duration and the physical environment all change the picture.
How smart devices and AI assistants add to everyday wireless signals
Smart speakers, phones, watches, Bluetooth earbuds, security cameras, connected appliances and Wi-Fi routers can create a busier mix of RF signals. AI makes many of these devices more useful, helping them recognise speech, automate routines or make predictions.
A smart home with several always-connected devices may have more wireless activity than a home with one router and a basic handset. That doesn't automatically mean harmful exposure. It means there are more sources worth understanding, especially if your router sits beside the sofa, your phone lives under your pillow, or every room has another connected gadget blinking away.
Wi-Fi, Bluetooth, 4G and 5G are wireless technologies that operate around you simultaneously. This can contribute to cumulative exposure, although the level varies with distance, transmission power and use. Our closer look at the electromagnetic soup in modern homes explores why that can feel like a lot, even when each device is doing something ordinary.
Why AI data centres create more EMF
Data centres use servers, power supplies, cooling systems and networking equipment. Their power quality depends on how these systems draw, convert and distribute electricity. Non-linear loads can produce harmonic distortion, grid harmonics and supraharmonics, which are conducted disturbances rather than a separate form of wireless transmission.
A large AI data centre may need transformers, switchgear, bus ducts, uninterruptible power supply (UPS) systems, heavy-duty feeders, substations and upgraded transmission or distribution connections. These systems must maintain power quality under demanding loads. UPS systems, converters and switching equipment can introduce harmonic distortion, supraharmonics and voltage transients.
An AI data center may also use filtering and power-factor correction to manage power quality. These measures can reduce harmonic distortion and supraharmonics, while poor performance can increase electrical pollution within the installation. Some consumers informally call this “dirty electricity”, but it is not a settled diagnosis or a single scientific measurement.
Transformers, busbars, feeders and substations can create EMF, particularly when carrying high currents. Cabling design, load balance and distance affect these fields and the site’s power quality. Harmonic distortion and supraharmonics may also contribute to electromagnetic interference, affecting nearby electronics.
The International Energy Agency estimates that data centres used around 415 TWh of electricity globally in 2024, and projects this could rise to roughly 945 TWh by 2030. AI is expected to be a major driver of that growth. More electricity demand can mean more or larger electrical infrastructure, particularly at and near large facilities.
AI data centers may therefore require more transformers, cabling and conversion equipment. Their power quality can vary with server demand, cooling loads and UPS operation. In this context, “dirty electricity” is best understood as an informal consumer term for measurable disturbances, including harmonic distortion and supraharmonics.
That doesn't necessarily mean a distant data centre automatically raises EMF levels across an entire neighbourhood. The strongest electromagnetic fields are usually close to electrical equipment and power routes. Conditions at a site boundary, in a nearby home or in a public space depend on the equipment layout, electrical load, cable design, distance, shielding and any new substations or transmission works.
A data center’s electrical pollution should therefore be assessed through appropriate measurements. Those checks can examine power quality, harmonic distortion and supraharmonics at relevant points. They can also identify whether “dirty electricity” is present on internal circuits, rather than assuming every AI data center creates the same conditions.
That being said, the potential impact needs robust scientific analysis. Measurement should distinguish conducted electrical disturbances from radiofrequency radiation and from low-frequency fields around power infrastructure.
A data centre can add local EMF sources, especially around its power systems. At the site boundary, it is crucial that we start measuring the conditions rather than assuming danger or safety.
IEA’s Energy and AI report provides the data-centre electricity figures. For technical context, IEC/TR 62271-208covers methods for assessing power-frequency electromagnetic fields from high-voltage switchgear and prefabricated substations.
AI, 5G and data centres affect different types of EMF exposure
RF fields come from wireless communication, including mobile networks, Wi-Fi and Bluetooth. This radiofrequency radiation is a form of non-ionizing radiation from wireless devices and infrastructure. Low-frequency electric and magnetic fields come from mains wiring, transformers, power supplies and electrical equipment.
AI services may push more data through 5G networks. Yet a base station on a building, a router on a shelf and a data-centre transformer are not interchangeable sources. Treating all EMF as one thing is a bit like calling drizzle, surf and steam the same because they all involve water.
What the current evidence says about 5G and public RF levels
More AI traffic does not currently establish a direct, measurable AI-driven rise in public 5G exposure. Readings vary with antenna design, traffic, proximity, walls, building layout and where you happen to be standing.
A review of measured mobile base-station RF levels found public exposures well below international reference levels in the settings assessed. A European policy brief on 5G exposure monitoring also reports levels well below safety limits.
Those limits are important guardrails. They are not a magic answer to every question about long-term exposure or individual symptoms. Science can keep testing, while people make sensible choices without getting dragged into panic.
Why data-centre electricity demand is rising so quickly
AI data centers rely on high-performance servers that run hot and draw far more power than ordinary office computing. That demand can affect power quality as well as total electricity use. Cooling is part of the bill, and AI data centers may require substantial power quality management.
Power-conversion equipment can produce harmonic distortion in electrical systems. These grid harmonics may affect power quality without automatically indicating a health risk. At higher frequencies, engineers may discuss supraharmonics, which are not simply another name for harmonic distortion.
Some commentators use dirty electricity or electrical pollution to describe unwanted electrical disturbances. Those terms are broad and are not interchangeable with power quality. They should not be treated as automatic evidence of harm. Dirty electricity can refer to different measurements, including harmonic distortion, voltage transients and other disturbances. Supraharmonics may also form part of that discussion, but the specific measurement matters.
Large hyperscale facilities can draw 100 MW or more, which is serious industrial-scale electricity demand. EMF may be present near high-current cabling, switchgear and transformers. Their effect on power quality, including supraharmonics and harmonic distortion, depends on the equipment and installation.
A nearby data center may be discussed in terms of dirty electricity, but that label alone tells you little about actual conditions. Power quality measurements can identify supraharmonics and other disturbances. Harmonic distortion may also affect equipment through electromagnetic interference, rather than directly affecting people.
You cannot look at a huge building, hear the word “AI” and know the field level at its boundary. You have to measure it. The same applies to power quality, harmonic distortion and supraharmonics, which require suitable instruments and careful interpretation.
What higher AI-related EMF exposure could mean for EHS individuals
Some people who identify with electromagnetic hypersensitivity, or EHS, report headaches, fatigue, poor sleep, skin sensations, tinnitus, dizziness, palpitations or trouble concentrating around wireless technology. Those symptoms are real experiences, even when the cause is unclear.
Blinded provocation studies have not shown a consistent ability for EHS individuals to detect EMF exposure. Symptoms can also relate to stress, sleep loss, vestibular conditions, screen time, medication, migraine and many other health factors. Persistent or worrying symptoms deserve a conversation with a GP, not a shrug.
Why the science remains difficult to interpret
Real life is messy. Signals change minute by minute, frequencies overlap, and nobody lives in a laboratory bubble. Cumulative exposure is difficult to characterise because sleep, work pressure, caffeine, anxiety, illness and bright screens can all affect how the body feels.
Safety standards focus on established heating effects. Research into possible non-thermal effects remains unsettled. A Public Health Ontario review of RF and 5G evidence found no clear human health risk from the newer bands studied under current exposure conditions, while also recognising the need for continued research.
EHS individuals may find symptoms difficult to separate from these overlapping influences. A measured exposure does not, by itself, establish a cause. Appropriate support should focus on reported symptoms and practical health needs.
Why smart cities may feel different from ordinary homes
In a dense urban area, mobile networks, public Wi-Fi, smart meters, transport systems, sensors and connected vehicles all share the same physical space. Wireless technologies can create more sources to assess, making the environment feel different from an ordinary home.
It can feel like a louder technological environment, even though RF isn't something your ears can hear. That feeling is worth taking seriously, but a higher signal count does not prove higher personal exposure.
EHS individuals may prefer clear information about local conditions. Still, local EMF levels need measuring rather than guessing. More signals in the air does not tell you the strength of any one signal where your body is.
Practical ways to reduce unnecessary EMF exposure from AI devices
You don't need to throw your phone in the sea and retreat to a cave. Aim for sensible exposure reduction, not total elimination. Create distance, reduce idle connectivity and notice how you feel over time.
Create lower-wireless spaces at home
Keep your router away from beds and places where you spend hours resting. Put Wi-Fi on a timer overnight if it suits your household. Use airplane mode when your phone doesn't need to connect, especially near your body.
Ethernet is a solid option for computers. These wired alternatives can also include headphones for long listening sessions. Turn off smart home features that nobody uses. Review privacy controls to limit unnecessary cloud connections.
Check app permissions, microphones, cameras and remote access on connected devices. Good privacy controls can reduce unwanted connections as well as protect your data. A “smart” toaster that never earns its keep is just clutter with an app.
Measure before changing your home or buying protection products
RF meters measure radiofrequency signals and can help compare readings at different distances. RF meters don't measure every type of field, so separate instruments may be needed for electric and magnetic fields.
Check each meter's frequency range, instructions and accuracy. Compare readings at different distances and times of day. Readings depend on the instrument, location, timing and nearby equipment.
Some people use the informal term dirty electricity for electrical disturbances on household wiring. It isn't a diagnosis, and a plug-in filter won't identify the source. Check power quality before buying filters, particularly if appliances switch loads regularly.
Appliances, chargers and dimmers can affect power quality. They may contribute to harmonic distortion, supraharmonics or other forms of electrical pollution. A specialist may investigate grid harmonics, harmonic distortion and voltage transients during a detailed assessment.
UPS and inverter equipment can also change power quality. If you use this equipment, include it when discussing dirty electricity with a professional. An assessment may look for supraharmonics and harmonic distortion as part of a wider review.
For persistent concerns, ask a qualified electrician to assess power quality and the installation. They can check whether the reported dirty electricity relates to wiring, appliances or supply conditions.
Don't treat one target number as a universal medical threshold. Cumulative exposure is also difficult to interpret from a single reading. Don't buy expensive shielding before you know what you are trying to reduce.
Use a sensible precautionary approach
Keep exposure as low as reasonably achievable without turning daily life into an anxious full-time project. Prioritise sleep, daylight, outdoor time and breaks from screens. These precautionary measures support good health practice, whatever your view on EMF.
Shielding materials may reduce some RF signals when properly designed and installed. They aren't a universal answer for low-frequency magnetic fields, electrical wiring or every source of interference.
Faraday fabric can work in some applications, but its performance depends on the weave, construction and grounding. Faraday fabric used in clothing is not automatically equivalent to Faraday fabric used for curtains or room treatments. Faraday fabric and other shielding materials should only be considered after identifying a specific source and taking suitable measurements.
People who describe electromagnetic hypersensitivity may experience genuine, distressing symptoms. EHS individuals deserve a supportive, non-judgemental approach, without assuming a diagnosis or dismissing their concerns. EHS individuals may benefit from reviewing sleep, stress, lighting, ventilation and other possible triggers with a qualified professional.
The Role of AV Edge
If you want to take an intentional approach to living with modern technology, AV Edge offers frequency-based wellness products designed for everyday use. Our wristbands and Protect Filters use proprietary EF Technology, which incorporates more than 300 frequency signatures. They are made for people who want to support a greater sense of balance and wellbeing while surrounded by modern connected technology, every day.
Many customers choose AV Edge alongside practical habits such as creating distance from devices, using wired connections where possible, switching off unnecessary wireless features and protecting their sleep routine. Explore the AV Edge human wellness and performance range to see which product may suit your routine.
AV Edge products and similar frequency-based products are not medical devices. They are not intended to diagnose, treat, cure or prevent any illness, and they should never replace evidence-based treatment or medical care.
Explore AV Edge Wellness Products
Frequently Asked Questions
Does AI itself give off electromagnetic radiation?
No. AI software does not emit EMF by itself. The data centres, servers, wireless networks, power supplies, networking hardware, and mobile devices used to run AI are responsible for producing RF or low-frequency fields.
Will AI make my home's EMF level higher?
It may if you add more connected devices or use wireless services more often, or live within close proximity to a data centre. Privacy controls can help limit unnecessary connections. The real change depends on device location, output and how close you are to them.
Are 5G networks used for AI below safety limits?
Available monitoring reports show public base-station exposure below international reference levels. Compliance provides useful context for day-to-day exposure, while research continues.
Can living near a data centre increase EMF exposure?
It can, particularly if new transformers, substations, switchgear or high-capacity cabling are installed nearby. For a data center, distance, equipment layout, electrical load and substations matter more than the building's label. Exposure outside the site depends on site design and shielding, so measurements are more useful than assumptions.
What is dirty electricity and other household electrical disturbances?
Terms such as dirty electricity, power quality and voltage transients describe technical or informal concerns about electrical supply. Electrical pollution is a non-standard consumer term, not a settled medical category. Identifying the source requires a site-specific assessment of wiring, appliances and incoming supply.
Can shielding reduce exposure from household devices?
Some shielding materials, including Faraday fabric, can reduce signals in suitable conditions. Performance depends on frequency, installation and grounding. Poorly fitted shielding may offer little benefit, so it should be assessed for the specific device and environment.
Can I measure EMF at home?
RF meters can indicate radiofrequency levels, but they have important limitations. Results depend on meter range, calibration, positioning and the type of signal being measured. A single reading cannot describe every source or exposure pattern.
What is the easiest way to reduce wireless exposure?
For practical exposure reduction, increase distance from transmitting devices and reduce close-contact time. Choose wired alternatives where convenient, and switch off wireless functions you don't need. These steps are practical precautions, not proof that ordinary consumer exposure causes illness.
Should I see a GP about symptoms I link to EMF?
Yes, if symptoms persist, are new or disrupt normal life. Some EHS individuals report headaches, fatigue, palpitations, dizziness or sleep issues, but these symptoms can have many causes. Speak to a GP, and seek urgent NHS help for severe chest pain, fainting, sudden weakness or serious breathing difficulty. Precautionary measures should not replace medical advice.
A Clearer Way to Live With AI
AI can add devices, increase data traffic and drive enormous electricity demand in data centres. Pragmatism should be the key direction. We should pressure authorities to start measuring the levels of EMF and allow individuals to take personal responsibility by creating distance, choosing wired options when they fit, and reducing unnecessary connectivity.
