Britain On Purpose

Campaign 6

Keep AI in Democratic Hands

Ensure the future of the most powerful AI systems is not decided by a handful of unelected companies and executives.

Artificial intelligence may become one of the most consequential technologies ever created. It could transform science, medicine, education, employment, defence, communications and the global economy.

That makes one question particularly important: who gets to decide how it develops?

Today, the most advanced AI systems require immense quantities of computing power, capital, specialist talent and data infrastructure. That naturally concentrates capability among a relatively small number of organisations.

Britain on Purpose believes democratic governments should take that concentration seriously — before, not after, these systems become embedded throughout society.

The campaign in 30 seconds

  • —The most powerful AI systems are being developed by a handful of private companies with limited democratic oversight.
  • —Good intentions inside those companies are not a substitute for independent evaluation, government expertise, and proportionate regulation.
  • —Britain needs world-class technical capability inside government — not just policy generalists trying to regulate technology they do not understand.
  • —Regulation should scale with capability: more power should bring more accountability, not less.

Democratic AI Accountability Scorecard

How well does Britain currently hold the most powerful AI systems to democratic account?

Independent evaluation of frontier AI systems before deployment
Government technical expertise sufficient to regulate effectively
Mandatory incident reporting for serious AI safety failures
Proportionate regulation that scales with capability and risk
Transparency about how AI is used in public-sector decisions
Meaningful domestic capability in AI research, compute and safety science

This is not an argument against AI companies

The people building advanced AI have achieved extraordinary things. Many are genuinely concerned about safety and responsible deployment. But that is not sufficient.

We do not design democratic institutions on the assumption that powerful people will always be wise. Good intentions are not governance.

Whether the number is five companies or twenty is ultimately less important than the underlying principle. A technology capable of reshaping society should not be governed almost entirely through private decisions made inside a handful of boardrooms.

What we want

Britain on Purpose calls on government to:

  1. 01Establish mandatory independent evaluation of the most capable AI systems before deployment, with results accessible to government and independent researchers.
  2. 02Build world-class technical AI expertise inside government — not just policy generalists, but researchers, engineers, security experts and evaluators.
  3. 03Require frontier AI developers to report serious safety and security incidents to relevant authorities promptly.
  4. 04Apply regulation proportionate to capability — lighter touch for modest applications, greater scrutiny where consequences could be large.
  5. 05Publish clear standards for how AI is used in public-sector decisions affecting individuals, with appropriate safeguards and rights of challenge.
  6. 06Maintain meaningful domestic capability in AI research, compute, safety science and evaluation — not total self-sufficiency, but enough to make genuine choices.

The pro-innovation case for democratic oversight

Strong democratic oversight is not the enemy of AI development. It is the foundation on which public trust is built.

Technologies that develop without adequate accountability tend to generate backlash. Proportionate, technically competent regulation creates the conditions for sustainable innovation — not a barrier to it.

Britain has an opportunity to demonstrate that democratic governance and frontier AI development are compatible. That is a competitive advantage, not a handicap.

What would change our mind

  • —Evidence that existing voluntary commitments by AI developers are producing outcomes equivalent to independent evaluation and mandatory reporting.
  • —Demonstrated government technical capability sufficient to evaluate and regulate frontier AI without additional investment.
  • —A credible international framework that makes domestic regulation unnecessary or counterproductive.

Sources and further reading

  1. 1.AI Safety Institute — GOV.UK
  2. 2.Frontier AI Taskforce Second Progress Report — GOV.UK
  3. 3.AI Regulation White Paper — GOV.UK

Democratic accountability and frontier AI development are not opposites. Britain can lead on both.