Why Artificial Intelligence Is Becoming an Early Fault Line in the 2028 Democratic Presidential Conversation

Artificial intelligence is moving from a specialist technology debate into mainstream American politics. Long before the 2028 presidential campaign formally begins, several prominent Democrats are already developing distinct approaches to AI regulation, data centres, jobs, public safety and international competition.

Artificial intelligence may still look like a technology policy issue, but in the United States it is rapidly becoming something much bigger. Questions about how quickly AI should develop, who should regulate it, who should pay for the infrastructure supporting it and how workers should be protected are increasingly becoming political questions.

That shift is beginning to influence the conversation surrounding the 2028 presidential election.

Several prominent Democrats who are frequently discussed as possible future presidential contenders have begun setting out positions on artificial intelligence. They include former vice-president Kamala Harris, California governor Gavin Newsom, Pennsylvania governor Josh Shapiro, Kentucky governor Andy Beshear, Arizona senator Mark Kelly, New York representative Alexandria Ocasio-Cortez, California representative Ro Khanna and former transportation secretary Pete Buttigieg. None of being discussed as a potential contender should be treated as confirmation of a 2028 candidacy.

What makes the emerging debate significant is that the politicians are not simply arguing over whether AI is "good" or "bad". Instead, several separate issues are converging: technological safety, unemployment, electricity demand, water consumption, corporate responsibility, national security, misinformation, privacy and America's technological competition with China.

The result is an unusually broad policy debate that could become one of the defining economic and technological questions of the next presidential cycle.

Why AI has suddenly become a political issue

For several years, political discussion around artificial intelligence largely concentrated on its potential economic advantages. Governments discussed innovation, productivity, scientific research and America's ability to compete technologically with China.

That conversation has not disappeared. However, public policy is increasingly focused on the costs and risks accompanying rapid AI development.

One concern involves the enormous physical infrastructure required to operate advanced AI systems. Generative AI depends heavily on large data centres filled with specialised computers. These facilities can consume substantial amounts of electricity and water and sometimes require new power infrastructure.

Consequently, debates about artificial intelligence increasingly occur not only in Silicon Valley or Washington but also in communities where companies want to construct data centres.

The Guardian reported on 26 September that polling had found considerable public unease: a New York Times/Siena survey found 61% opposition to data-centre construction among voters surveyed, while a CNN/SSRS poll found 71% saying government was not doing enough to regulate AI development.

The concern is also increasingly bipartisan at the state level. In September, for example, Republican Texas governor Greg Abbott halted new state-issued data-centre permits while officials examined their effects on the electricity grid, illustrating that infrastructure concerns are not confined to Democratic-led states.

At the same time, workers are asking a different question: what happens if increasingly capable systems automate substantial portions of existing employment?

The answer is uncertain. AI could create occupations and increase productivity in some industries while reducing demand for certain kinds of work in others. That uncertainty is pushing politicians to think beyond traditional technology regulation and toward labour-market policies, education, retraining and social protection.

These concerns explain why AI is increasingly being discussed as an economic policy issue rather than simply a technology issue.

Kamala Harris: regulating AI while preserving its benefits

Kamala Harris enters the emerging debate with experience from the Biden administration.

As vice-president, Harris was closely involved in the administration's discussions with technology companies over artificial-intelligence safety. A 2023 White House meeting involving Harris and executives from major AI companies addressed risks involving safety, national security, privacy, employment, civil rights and democratic institutions.

Her position at the time emphasised that policymakers should consider existing AI harms alongside longer-term catastrophic risks. During the 2023 international AI safety discussions, she highlighted concerns involving misinformation, discrimination, privacy and manipulation while arguing that technological innovation and public protection should not be treated as mutually exclusive goals.

Her position has subsequently moved toward stronger intervention.

On 14 September 2026, Harris publicly supported slowing the development of the most advanced frontier AI systems. She also called for new federal legislation, an oversight institution capable of independently testing advanced systems and international coordination that would include countries such as China.

That is important because it moves beyond conventional consumer protections.

A policy of slowing frontier-model development would raise difficult questions about exactly what constitutes a sufficiently powerful system to require restrictions, who conducts safety evaluations and whether American companies could slow development without similar commitments from foreign competitors.

Harris's approach therefore connects domestic regulation to international governance.

Gavin Newsom: California as an AI regulatory laboratory

California governor Gavin Newsom faces a particularly complicated political environment because his state contains many of the world's leading technology companies.

California therefore has strong economic incentives to encourage AI innovation while also facing pressure to regulate an industry concentrated within its borders.

Newsom has attempted to pursue both objectives.

California has promoted the use and development of artificial intelligence, including deploying AI tools within state government. In September 2026, for example, the state introduced AskCA, an AI-powered system designed to help Californians navigate government services.

But Newsom has simultaneously expanded safety regulation.

On 18 September 2026, he issued an executive order accelerating implementation of independent oversight of frontier AI developers. The order directed experts to consider stronger third-party verification of company safety systems and mechanisms capable of shutting down advanced systems under specified emergency conditions.

This "kill switch" concept is one of the more technically ambitious proposals emerging in current AI regulation.

In practice, it raises complex questions. Who determines that a model has become dangerous enough to deactivate? How would developers ensure the shutdown mechanism itself could not be circumvented? Would such requirements apply only to the largest frontier systems, or eventually extend to less advanced models?

Newsom's approach illustrates one emerging Democratic position: permit AI development but require increasingly sophisticated safety mechanisms as capabilities grow.

Josh Shapiro: innovation combined with enforceable rules

Pennsylvania governor Josh Shapiro has positioned himself differently.

Rather than presenting AI primarily as a technological threat, Shapiro has argued that its benefits can be retained while governments impose stronger safeguards.

In a September 2026 speech at the AI Horizons summit in Pittsburgh, Shapiro said the technology could improve areas including healthcare and workplace productivity but argued that powerful AI systems should operate within a federal regulatory framework involving independent oversight, greater transparency and enforceable safety standards.

His analogy was straightforward: industries such as aviation and pharmaceuticals operate under safety regulations before products can be widely deployed, and he argues that highly consequential AI should face comparable oversight.

Pennsylvania has also acted on data centres.

On 18 August 2026, Shapiro signed an executive order removing AI data-centre proposals from the state's fast-track permitting process. Developers seeking permits must meet requirements dealing with electricity affordability, environmental protection, employment, transparency and local community involvement.

The approach reflects a broader distinction emerging in Democratic AI policy.

Some politicians focus primarily on whether the AI models themselves are safe. Shapiro's policies address both the digital technology and the physical infrastructure required to operate it.

That means questions about AI become questions about electricity prices, environmental permitting, employment and local government power.

Andy Beshear: protecting communities from data-centre costs

Kentucky governor Andy Beshear has focused particularly heavily on those infrastructure effects.

In August 2026, Beshear issued an executive order establishing conditions for data-centre projects seeking to operate in Kentucky.

Developers are expected to demonstrate that their projects will not shift electricity costs onto existing households and businesses. They must submit energy plans explaining how their power requirements will be met, while state environmental authorities are directed to examine possible effects on air quality, water availability, wetlands and other natural resources.

Beshear's position is not a blanket prohibition on data centres.

Local communities retain an important role in determining whether projects proceed. The policy attempts instead to make developers responsible for demonstrating that economic investment will not impose unacceptable costs on surrounding communities.

However, the policy has also generated legal and administrative questions. Critics have questioned how much authority an executive order can exercise over an independent utility regulator without supporting legislation from the state legislature.

That disagreement illustrates another problem future presidents will encounter: even when there is political agreement that AI infrastructure requires oversight, designing legally durable regulatory mechanisms may be difficult.

Mark Kelly: make the AI industry help finance the transition

Arizona senator Mark Kelly has concentrated heavily on the economic consequences of AI, particularly its effect on workers.

In September 2025, Kelly proposed an "AI Horizon Fund" financed through contributions associated with major AI companies. His framework proposed using the money for workforce training, energy infrastructure and support for communities adapting to technological disruption.

The proposal moved from policy paper toward legislation in September 2026.

On 24 September, Kelly introduced the Make AI Work for Americans Act. According to his office, the legislation would establish a federal trust fund supported through taxes associated with excess profits, digital advertising and AI usage. The money would support worker training, entrepreneurship, unemployment protection and other programmes intended to assist workers affected by technological change.

Kelly's proposal therefore asks one of the most consequential questions in AI politics: who should pay for economic adaptation?

Governments have traditionally financed unemployment programmes, education and infrastructure primarily through general taxation. Kelly's approach would place greater responsibility specifically on companies benefiting economically from artificial intelligence.

Whether such a funding mechanism could pass Congress is a separate political question, but the underlying issue is likely to persist even if the proposal itself does not.

If AI produces major productivity gains while simultaneously displacing some workers, policymakers will have to decide how the benefits and costs of that transition are distributed.

Alexandria Ocasio-Cortez: slowing infrastructure expansion until rules exist

Representative Alexandria Ocasio-Cortez has adopted one of the more restrictive positions regarding AI infrastructure.

In June 2026, she introduced a House version of the Artificial Intelligence Data Center Moratorium Act alongside a Senate initiative led by Bernie Sanders.

The legislation would temporarily stop construction of new AI data centres and expansion of existing facilities until Congress enacted comprehensive legislation dealing with AI's economic, environmental and safety consequences.

Her concerns extend beyond infrastructure.

Ocasio-Cortez has raised issues involving surveillance, sexually explicit AI-generated deepfakes, electricity costs and possible employment disruption.

Her approach represents a fundamentally different regulatory philosophy from politicians who favour allowing development while introducing safeguards simultaneously.

A moratorium reverses the sequence: establish rules first, then permit further infrastructure expansion.

That distinction could become increasingly important. As demand for computing capacity grows, policymakers will have to decide whether regulation should attempt to keep pace with expansion or whether some forms of development should temporarily stop while governments establish clearer rules.

Ro Khanna: regulation through international cooperation

Representative Ro Khanna occupies another distinctive position because he represents Silicon Valley while also advocating stronger AI safeguards.

His approach increasingly links domestic AI regulation with American foreign policy toward China.

Khanna has argued for agreements between Washington and Beijing involving safety standards, inspections and mechanisms for managing extremely advanced AI systems.

In proposals reported in September, Khanna supported international restrictions on recursive self-improvement and superintelligence, model audits before release and coordinated shutdown mechanisms for advanced systems. He also argued that an international inspection framework could help verify compliance.

At the domestic level, Khanna has also proposed a Data Center Bill of Rights designed to give communities greater control over where data centres are constructed and how their environmental and public-health effects are managed.

His approach highlights a central dilemma in AI governance.

Artificial intelligence companies operate internationally, their models can often be accessed across borders and technological competition is closely connected to national security. Consequently, regulation confined to individual American states may be insufficient for the most advanced systems.

Yet reaching enforceable international agreements, particularly between strategic competitors such as the United States and China, presents considerable diplomatic and verification challenges.

Pete Buttigieg: AI as a transformation of society

Former transportation secretary Pete Buttigieg has framed the issue more broadly.

In a June 2025 essay, Buttigieg argued that American society and its political institutions were underprepared for the scale of change artificial intelligence might bring.

His concerns extended beyond immediate technological safety to questions involving employment, wealth distribution, political power, social isolation and citizenship.

This perspective treats AI less as another industry requiring regulation and more as a general-purpose technology capable of transforming multiple parts of society simultaneously.

Historical comparisons are useful here, although imperfect. Technologies such as electricity, automobiles and the internet eventually influenced employment, infrastructure, law, communication and culture far beyond their original applications.

If artificial intelligence follows a comparable trajectory, then regulating individual products may not be sufficient. Governments may need to reconsider education systems, labour protections, intellectual-property rules, competition law and social-security programmes.

That broader framing could become increasingly important as politicians attempt to explain what AI policy means to voters who may never directly interact with frontier-model development but may nevertheless experience changes in their workplace or community.

The deeper Democratic disagreement

Taken together, these positions reveal that there is not yet one unified Democratic approach to artificial intelligence.

There is considerable overlap. Most of the figures discussed here support some form of stronger oversight, and several have expressed concern about workers, data centres or advanced-model safety. Recent reporting describes Democratic politicians as increasingly responding to public concern over AI while the current Republican administration has generally emphasised American technological leadership and has been more sceptical of extensive regulation.

But important differences remain.

One question is how quickly AI should continue developing.

Harris has supported slowing frontier-model development. Ocasio-Cortez supports temporarily stopping new data-centre construction until broader federal rules exist. Newsom is allowing development while constructing stronger safety mechanisms around it. Shapiro similarly argues that innovation and regulation can proceed together.

Another difference concerns economic policy.

Kelly focuses on making companies benefiting from AI contribute financially toward worker adjustment and infrastructure. Beshear and Shapiro concentrate more directly on preventing data-centre developers from transferring electricity, environmental or infrastructure costs onto local residents.

Khanna emphasises international agreements alongside domestic regulation, while Buttigieg has framed AI as a wider social transformation requiring political imagination extending beyond technical regulation.

These differences could matter significantly if AI becomes a major issue during the 2028 Democratic primary.

Jobs may ultimately become the most important political question

Catastrophic AI risks attract substantial attention because their potential consequences are enormous. But employment may become the issue that connects AI most directly with everyday electoral politics.

Millions of American workers perform activities that AI systems can increasingly assist with or partially automate: writing, translation, software development, administration, customer service, research, financial analysis and design.

Automation does not necessarily eliminate an occupation simply because technology can perform some of its tasks. Jobs frequently change rather than disappear completely.

However, even partial automation can reduce the number of workers required to produce the same output.

That possibility transforms AI from an abstract debate about advanced technology into a question of wages, employment security and economic inequality.

Kelly's worker fund is one attempt to address that possibility. Buttigieg has similarly emphasised questions surrounding work and inequality, while Khanna has asked technology companies how they intend to prevent mass unemployment.

Future candidates are therefore likely to face increasingly specific questions: Should workers displaced by AI receive wage insurance? Should companies finance retraining? Should automation gains be taxed differently? Should labour law give workers greater influence over how AI is introduced in workplaces?

Those debates could eventually matter more politically than technical discussions about model architecture.

Data centres connect AI policy to ordinary household economics

Data centres may prove equally important because they make artificial intelligence physically visible.

An AI chatbot appears intangible. A hyperscale data centre is not.

It requires land, power connections, water, construction permits and sometimes substantial public incentives. Residents may therefore encounter AI policy through electricity bills, water concerns, tax arrangements or proposed developments in their communities.

This helps explain why governors such as Newsom, Shapiro and Beshear are dealing with AI not only through technology regulation but through energy and environmental policy.

It also helps explain Ocasio-Cortez's call for a temporary construction moratorium.

The emerging political question is no longer simply whether America should build enough computing infrastructure to remain technologically competitive.

It is also who bears the cost of building it.

China complicates almost every regulatory proposal

Any American attempt to slow advanced AI development immediately encounters another argument: what happens if China does not slow down?

This is why several Democratic proposals combine regulation with international cooperation rather than treating AI safety purely as a domestic matter.

Shapiro has called for international collaboration while also arguing that the United States should maintain technological leadership. Khanna has proposed direct U.S.–China agreements covering advanced-model safety and verification. Harris has likewise supported an international framework involving major AI powers.

The problem resembles other areas of strategic technology policy.

A country may decide that certain restrictions improve safety, but policymakers also have to consider whether unilateral restrictions could shift technological leadership elsewhere.

Conversely, unrestricted competition can create incentives for companies and governments to move faster than they otherwise would because each fears being overtaken.

This tension between competition and coordination is likely to remain central regardless of which party controls Washington.

Why the 2028 AI debate matters

The significance of the current discussion is not that the 2028 campaign has already begun in any formal sense.

It has not.

Rather, politicians who may play important roles in the next presidential cycle are beginning to develop recognisably different policy frameworks before the campaign becomes dominated by electoral messaging.

That makes the current period particularly important.

The fundamental questions are still open.

How powerful should an AI system become before independent testing is mandatory? When should government be able to stop a model from operating? Should communities be able to reject data centres?

Should AI companies finance worker retraining? Should new infrastructure be paused until federal legislation exists?

Can the United States regulate aggressively without surrendering technological advantages to China?

And can Washington negotiate meaningful international AI rules that can actually be verified?

There is no settled Democratic answer to those questions.

Instead, an emerging spectrum can already be seen: from stronger oversight combined with continued technological expansion, to temporary restrictions designed to give regulation time to catch up, to international agreements intended to prevent uncontrolled competition.

That policy debate may be more consequential than the political personalities involved.

Artificial intelligence is increasingly touching employment, energy, national security, education, privacy, misinformation and economic inequality simultaneously. Few issues confronting the next American president are likely to remain confined to a single government department.

For that reason, AI policy is moving from the margins of political debate toward its centre.

By 2028, presidential candidates may no longer be able to speak about artificial intelligence simply by promising either "innovation" or "regulation". Voters, workers, companies and local communities are increasingly demanding answers to more specific questions: what kinds of AI should be permitted, under what rules, who benefits economically, who absorbs the risks and who ultimately has the authority to say that technological development has gone too far.

Those questions are already beginning to shape the politics of the next presidential era.