AI Superintelligence Boom | DISABLED ENTREPRENEUR

Written by Editor

Disclaimer: This article is for informational, educational and public-interest purposes only. It does not constitute legal, environmental, medical or financial advice, nor does it endorse any political party or politician. Political proposals and legislation may change as they progress through the relevant legislative processes. Readers concerned about the use of their personal or health information should consult the relevant organisation’s privacy notice or seek independent data-protection advice.

Pause Before We Power the AI Boom: Bernie Sanders’ Data Centre Moratorium, Disability, Privacy and the Environment

Should governments temporarily slow the construction of enormous AI data centres while laws, environmental protections and human rights safeguards catch up with the technology?

Artificial intelligence is frequently described as one of the most transformative technologies since the internet, but behind every chatbot, image generator, automated medical system and increasingly sophisticated AI model lies something far less visible: physical infrastructure.

AI does not exist in the clouds in any literal sense.

It operates through enormous data centres containing servers, networking equipment, cooling systems and other infrastructure requiring electricity, land, water and complex supply chains.

As governments race to attract AI investment, a growing political debate is asking whether infrastructure is being constructed more quickly than democratic institutions can establish the rules governing it.

US Senator Bernie Sanders has become one of the most prominent politicians calling for a temporary halt.

On 25 March 2026, Sanders and Representative Alexandria Ocasio-Cortez announced the Artificial Intelligence Data Centre Moratorium Act, proposing an immediate federal moratorium on new AI data centres and expansions until national safeguards addressing safety, workers, communities, utility costs and the environment are established.

Sanders has subsequently gone further. On 3 September 2026, he and Representative Greg Casar announced separate proposed legislation, the Ban Artificial Superintelligence Act, which would permanently prohibit artificial superintelligence while temporarily pausing certain advanced AI development until federal safety rules were established.

Then, on 15 September, Sanders called for the United States and China to negotiate an agreement to pause AI development, arguing that governments need more time to understand and regulate technology developing at extraordinary speed.

These are separate proposals, but they share one fundamental question:

Should humanity be building increasingly powerful AI infrastructure faster than governments can understand its consequences?

What Is a Moratorium?

A moratorium is a temporary suspension, pause or prohibition on an activity.

It does not necessarily mean something is permanently banned.

Governments sometimes introduce moratoriums when technology, infrastructure or commercial activity is developing faster than regulation. The purpose can be to provide time for research, consultation, legislation, environmental assessment or safety standards to be established before further expansion occurs.

Under the Sanders, Ocasio Cortez proposal, the intention is not simply to switch off existing artificial intelligence. Their proposed data-centre moratorium would temporarily stop further expansion while Congress establishes comprehensive safeguards.

According to Sanders’ announcement, those safeguards would address AI safety, workers’ ability to share in economic gains, privacy and civil rights, electricity and utility prices, community impacts and environmental protection.

Supporters therefore describe the proposal as applying the precautionary principle: when an activity could create serious or irreversible consequences, society considers those consequences before allowing expansion at unlimited speed.

Critics, however, argue that restricting infrastructure could slow innovation, discourage investment and disadvantage countries that pause development while competitors continue.

That disagreement is becoming one of the defining policy questions of the AI age.

Data Centres Are Not Environmentally Invisible

Every question asked of a large AI system ultimately requires physical computation somewhere.

The infrastructure may be hundreds or thousands of miles from the person using the software, but its environmental footprint remains real.

Data centres can create substantial demands for electricity. Cooling infrastructure may require significant quantities of water, depending on design, climate and technology. Construction requires land and materials, while backup generators can create additional emissions.

Hardware also has a lifespan.

Servers, processors, networking equipment and cooling technology can become obsolete rapidly as AI companies compete to install increasingly powerful chips. This raises another environmental concern: electronic waste.

In Britain, the government’s own AI Growth Zones programme illustrates the extraordinary scale involved. Applicants are expected to demonstrate access to at least 500MW of power capacity by 2030, as well as sufficient water availability and discharge capacity to support AI infrastructure on that scale.

That does not mean every data centre consumes its full theoretical capacity continuously, but it demonstrates why energy planning, water resources and local infrastructure have become central to the debate.

There are also economic questions about who pays when electricity networks need major upgrades.

In the United States, these concerns have moved beyond one political faction. On 16 September 2026, the US House of Representatives passed the Ratepayer Protection Act by 417 votes to 3. The measure requires state utility regulators to consider whether very large electricity users such as data centres should bear additional infrastructure costs rather than those costs being shifted onto ordinary households.

The issue is therefore not simply whether AI is useful.

It is also who pays for the physical infrastructure that makes AI possible.

Why This Matters Particularly to Disabled People

Artificial intelligence has enormous potential to improve accessibility.

AI-powered speech recognition can generate captions and transcripts. Computer vision may help blind or visually impaired people interpret their surroundings. Voice technology can assist people who cannot easily use keyboards or touchscreens. AI could improve accessible public services, assist with communication, reduce administrative barriers and potentially contribute to medical research and earlier diagnosis.

The UK Government itself acknowledged in July 2026 that AI-powered transcription and translation technologies have potential to improve accessibility for Deaf and hard-of-hearing people and others experiencing communication barriers.

Consequently, disabled people have a significant interest in ensuring that valuable AI innovation continues.

Yet disabled people may also experience some of the consequences of poorly planned infrastructure more severely.

Energy is a prime example.

Some disabled people rely upon electrically powered wheelchairs, hoists, ventilators, pressure-relieving equipment, refrigeration for medication, charging equipment or additional heating because of their health conditions.

The Energy Saving Trust notes that disability-related costs can include powering electric wheelchairs and hoists and having heating switched on more frequently or for longer.

Government-commissioned research published in 2025 also found that energy, electricity and utility bills were the area in which disabled respondents most frequently reported impairment-related increases in costs.

This creates an important equality question.

If enormous commercial data centres increase pressure on electricity infrastructure or contribute to additional network costs, safeguards may be needed to ensure that households, particularly those with unavoidable medical energy requirements, are not expected to subsidise infrastructure developed for some of the world’s largest technology companies.

The same principle applies to water, air quality and local development.

Technological progress should be examined not only through the question “Can we build it?”, but also through questions about who benefits, who carries the cost and whether vulnerable communities are adequately protected.

Scotland Has Already Entered the Debate

This is no longer purely an American discussion.

On 16 September 2026, the Scottish Parliament debated a motion calling for a moratorium on planning applications for new hyperscale data centres until Scotland had produced a national strategy and updated planning guidance.

The original blanket moratorium wording was amended. The final motion, passed by 80 votes to 26 with 15 abstentions, instead called for robust environmental standards, community consideration and national planning guidance, and stated that no planning or consenting decisions on such applications should be made until that work is completed.

Separately, the Scottish Government has introduced a direction requiring an Environmental Impact Assessment for new data centres exceeding 50MW. The requirement came into force on 17 September 2026.

The Scottish Government specifically recognised that AI creates national considerations involving energy use, water consumption and hardware requirements.

Scotland has therefore not adopted Sanders’ exact federal proposal, but it has moved towards a precautionary approach in which planning cannot simply race ahead without examining environmental and community consequences.

The debate now becomes whether England, Wales and Northern Ireland should consider comparable safeguards, whether existing planning and environmental systems are sufficient, or whether a UK-wide framework would provide greater consistency.

The Argument for a Wider UK Pause

Those supporting a temporary pause could argue that government needs time to understand cumulative demand rather than examining every development as if it existed in isolation.

One data centre may be manageable.

Dozens of hyperscale developments competing simultaneously for electricity connections, water, land and infrastructure create a different policy problem.

A temporary pause could allow governments to calculate anticipated energy and water requirements, establish environmental standards, decide who should finance necessary grid upgrades, examine the effect on household bills, require meaningful community consultation and determine what constitutes an environmentally responsible data centre.

It could also create time for Parliament to examine questions surrounding employment, surveillance, copyright, automated decision-making, cybersecurity, human control and the increasing concentration of technological infrastructure within a relatively small number of multinational companies.

Opponents would respond that data centres also support banking, healthcare, communications, public services, research and everyday digital services — not simply generative AI.

The UK Government designated data centres as critical national infrastructure in 2024, reflecting their importance to services ranging from patient records to financial systems.

The government also views domestic data-centre capacity as strategically important for Britain’s economy, resilience and AI ambitions.

There is therefore a genuine policy trade-off between precaution and investment.

Possible alternatives to a blanket moratorium include stricter environmental assessments, mandatory renewable-energy requirements, water limits, community-benefit agreements, transparency about electricity consumption, requirements for waste-heat reuse and rules preventing households from carrying infrastructure costs generated by large commercial users.

AI, Human Rights and the Question of Privacy

Energy is only one part of this debate.

Artificial intelligence depends upon data.

Depending on the application, systems may interact with photographs, voices, search histories, purchasing behaviour, location information, employment records, health information and enormous collections of publicly available material.

Not every use of personal information represents a privacy infringement. Organisations can lawfully process data under UK data-protection legislation where an appropriate legal basis exists.

Nevertheless, the scale of modern data processing raises legitimate questions about transparency, necessity, proportionality, security and meaningful public understanding.

People increasingly want to know:

Who holds my information?
Why is it being processed?
Who can access it?
How long is it retained?
Can it be transferred elsewhere?
Can it be used to train AI?
Can I object?
And what happens if something goes wrong?

Those questions become especially important when the information concerns health.

The NHS, Palantir and Patient Data

The American-founded technology company you may have heard about in connection with NHS data is Palantir Technologies.

Palantir Technologies UK Ltd supplies the technology behind part of NHS England’s Federated Data Platform (FDP).

It is important to clarify that this concerns NHS England and should not automatically be described as applying to NHS Wales, NHS Scotland or Health and Social Care Northern Ireland.

It is also important to distinguish between Palantir having a contractual role processing data, and Palantir owning NHS patient records.

According to NHS England, Palantir operates as a data processor under data-protection law and must process information according to NHS instructions. NHS England states that Palantir cannot commercialise or market NHS data and cannot use NHS data to develop its own products or train its own AI models. Data in the FDP is stored in the United Kingdom.

However, scrutiny of access arrangements has continued.

NHS England confirmed in 2026 that three Palantir engineers had administrative-level access to the National Data Integration Tenant to provide technical support, while additional supplier engineers had more limited project-specific access. NHS England says this access is tightly controlled and that it remains the data controller.

More recently, reports have also highlighted a pilot involving data from five GP practices in Cheshire and Merseyside being shared through local use of the Federated Data Platform, prompting renewed debate about transparency and public trust. NHS officials have said this is a local trial operating under data-sharing arrangements rather than incorporation of GP data into the national platform.

For patients, the distinction matters.

A technology company providing infrastructure under a processor contract is not legally equivalent to a company being handed NHS records to use however it wishes.

At the same time, members of the public are entitled to ask detailed questions about access controls, cybersecurity, contractor involvement, audit trails, data minimisation and whether the arrangements remain proportionate.

Transparency becomes particularly important because trust is fundamental to healthcare.

People must be able to seek medical treatment without fearing that deeply personal information will subsequently be used for unrelated commercial or technological purposes.

Can Patients Stop Their NHS Data Being Shared?

People in England have several mechanisms for exercising data-protection choices, although no single opt-out prevents every lawful use of NHS information.

The main options include:

National Data Opt-Out: allows people to prevent confidential patient information from being used for many research and planning purposes. It can be set or changed at any time. However, NHS guidance specifically states that this does not automatically apply to the Federated Data Platform, particularly where information is being processed for direct care or another exempt purpose.
Type 1 Opt-Out: patients can ask their GP practice not to share their GP data for certain research and planning purposes. This does not prevent all data processing elsewhere in the NHS.
Local opt-outs: some NHS trusts or Integrated Care Boards provide additional local choices, depending upon the system and purpose involved.
Data-protection rights: depending upon the lawful basis and circumstances, people may have rights to access their information, request correction, seek restriction or erasure in certain situations, and object to some forms of processing. These rights are not absolute in every circumstance.
Raise a complaint: concerns can initially be raised with the organisation processing the information or its Data Protection Officer. If someone believes data-protection law has been breached, they can also raise the matter with the Information Commissioner’s Office.

This distinction is important because “I do not consent” does not necessarily mean that every NHS organisation must stop processing information.

Public bodies may sometimes process personal information without consent where another lawful basis applies, for example, providing healthcare, complying with legislation or performing a defined public task.

The central issue is therefore whether processing is lawful, necessary, proportionate, secure and transparent.

Disability Data Requires Particular Care

Health and disability information can reveal some of the most intimate details about a person’s life.

Data may identify diagnoses, medication, mobility restrictions, mental health conditions, communication needs, cognitive impairments and other information that could potentially expose someone to discrimination if mishandled.

As AI becomes increasingly involved in employment, insurance, finance, healthcare and government administration, safeguards against discriminatory automated decision-making become increasingly important.

AI should not create a future in which people with complex health conditions are reduced to datasets and risk scores without meaningful human consideration.

At the same time, responsible use of health data can produce enormous public benefit.

Research using properly governed medical data may help scientists identify disease patterns, develop treatments, improve hospital capacity and understand inequalities.

The answer is therefore not necessarily to prohibit health-data research.

It is to ensure that the people whose lives are represented within datasets remain protected by law, oversight, security and accountability.

AI Could Transform Disability, If Disabled People Are Included

There is another danger in slowing AI indiscriminately.

Some of the people who may benefit most from artificial intelligence are disabled people.

A person unable to type easily may use voice-controlled AI.
Someone with dyslexia may use AI to structure or proofread text.
A visually impaired person may use computer vision to describe an image.
Someone with hearing loss may rely upon automated captioning.
A disabled entrepreneur who cannot easily travel may use AI to perform research, administration and communication from home.

Medical AI may eventually contribute to treatments for neurological conditions, genetic disorders, paralysis and other conditions currently considered difficult or impossible to treat.

This means disability policy cannot simply be pro-AI or anti-AI.

The more meaningful question is what kind of AI society wants.

Innovation that improves accessibility while respecting privacy, human rights and environmental sustainability is very different from uncontrolled technological expansion in which communities have little information about who is using their resources or personal data.

Jobs, Automation and Disabled Workers

Sanders has also repeatedly connected AI regulation with employment.

He argues that rapid automation could eliminate substantial numbers of jobs and concentrate economic benefits among technology owners.

Predictions about the precise scale of future job losses vary enormously, and nobody can know with certainty which occupations will disappear, which will change and which entirely new jobs will be created.

Nevertheless, disabled workers deserve particular consideration.

Remote employment and digital work have opened opportunities for people who may experience barriers travelling to conventional workplaces.

AI could make those opportunities more accessible.

Conversely, if employers use AI primarily to reduce staffing, automate administrative work and increase productivity without considering equality, some disabled workers could find themselves competing in a shrinking pool of accessible employment.

Human oversight will also remain important where AI is used for recruitment, performance monitoring, benefits, healthcare or workplace decision-making.

Efficiency should not automatically override equality.

The Bigger Question: Who Controls the AI Revolution?

The debate surrounding data centres ultimately reveals something much larger.

Artificial intelligence is not merely a new consumer technology.

Governments are making decisions today about energy infrastructure, planning law, public-sector data, national security, education, employment and healthcare that could shape society for decades.

Sanders’ position is that those decisions should not effectively be made by a small number of technology corporations.

Others argue that excessively restrictive regulation would allow rival countries to overtake Western economies, weaken innovation and delay technologies capable of delivering substantial social benefits.

Both perspectives raise legitimate issues for democratic scrutiny.

The debate therefore should not be reduced to whether somebody “likes AI”.

People can recognise AI’s extraordinary potential while questioning unlimited data-centre construction.

They can support medical AI while demanding stronger health-data safeguards.

They can use ChatGPT while asking how the servers supporting it are powered.

They can welcome accessibility technology while questioning whether disabled households should face higher infrastructure costs.

Technology and accountability do not have to be opposites.

Conclusion

Bernie Sanders’ call for an AI data-centre moratorium has moved a previously niche question into mainstream political discussion.

The United States is debating electricity costs and AI regulation. Scotland has introduced mandatory environmental assessment requirements for large data centres and, following its 16 September parliamentary debate, has called for planning decisions to wait while national guidance is developed. Britain, meanwhile, continues pursuing substantial AI infrastructure through its AI Growth Zone programme.

The policy options range from allowing expansion under existing rules, through stronger environmental and consumer safeguards, to temporary construction pauses while comprehensive frameworks are established.

For disabled people, there are substantial interests on both sides.

AI could deliver accessibility, independence, employment opportunities and medical breakthroughs. Yet disabled households may be particularly exposed to rising energy costs, discriminatory automated systems, inaccessible digital transformation and misuse of highly sensitive health information.

Perhaps the most important question is therefore not whether humanity should embrace or reject artificial intelligence.

It is whether technological development can take place while preserving human rights, privacy, environmental sustainability, affordability, accessibility and democratic accountability.

The servers may be artificial intelligence’s engine room.

But the decisions governing them remain human ones.