Who regulates AI, what governments and companies can actually control, and where accountability sits when automated systems cause harm.
AI governance is not one switch labelled safe or unsafe. Real control is distributed across developers, deployers, regulators, courts, technical standards, infrastructure providers and users. These guides map who can do what, what current UK rules actually cover, and where the gaps remain.
Audit whether an organisation has the basic ownership, access, logging, approval and shutdown controls around an AI deployment.
Pick the question closest to yours. Each guide gives you the answer first, then the evidence, the important caveats and practical next steps.
WHO SETS THE RULES?The UK does not currently have one all-purpose AI regulator. Existing regulators apply laws and sector rules within their remits, while government policy, technical standards and newer online-safety duties shape how AI is developed and used.
Read guide →
WHO SETS THE RULES?Yes. AI activity can already fall under data protection, consumer, equality, product, financial-services, online-safety and other laws. What the UK lacks is a single horizontal statute equivalent to one universal AI rulebook covering every use.
Read guide →
WHO SETS THE RULES?Governments can regulate companies, restrict services, enforce laws, control some infrastructure and impose sanctions or licensing conditions. They cannot literally erase every copy of software worldwide with one order, so “shut down AI” is a governance and enforcement problem, not a single technical button.
Read guide →
WHO SETS THE RULES?The UK AI Safety Institute is a government-backed technical research and evaluation body focused on advanced AI safety. It studies capabilities and risks and develops evaluation methods; it is not a general consumer regulator with a power to fine every AI company.
Read guide →
WHO SETS THE RULES?Responsibility normally attaches to people and organisations — developers, deployers, employers, service providers or other actors — through existing legal duties. An AI system does not become a convenient legal person that automatically absorbs everyone else’s responsibility.
Read guide →
WHO SETS THE RULES?Governance reduces risk by limiting access, requiring evaluations, assigning accountability, creating incident-response powers and making high-impact deployments auditable. None of those controls proves catastrophe is impossible; they reduce the number of unchecked pathways from capability to consequence.
Read guide →We label controlled evaluations, real-world incidents, expert forecasts and speculation differently. Read the methodology before treating any probability or timeline as a fact.