The UK is trying to figure out AI regulation, and it’s not going smoothly. After years of positioning itself as a “pro-innovation” alternative to the EU’s strict approach, the UK is now moving toward more structured regulation — and the details matter for anyone building or deploying AI in Britain.
What’s Happening
The UK’s approach to AI regulation has gone through several phases:
Phase 1 (2023-2024): Light touch. The previous Conservative government published a “pro-innovation” AI regulation framework that relied on existing regulators (FCA, Ofcom, CMA, etc.) to handle AI in their domains. No new AI-specific legislation. The idea: let regulators adapt existing rules rather than creating new ones.
Phase 2 (2025): The shift. The Labour government, elected in 2024, signaled a move toward more structured regulation. The AI Safety Institute (established after the Bletchley Summit) was expanded and given more authority. New proposals for AI regulation began circulating.
Phase 3 (2026): Implementation. The UK is now developing concrete AI regulatory proposals. The approach is more structured than the previous government’s but less thorough than the EU AI Act. It’s a middle ground that tries to balance innovation with safety.
The Key Proposals
Mandatory reporting for frontier AI models. Companies developing the most powerful AI models would be required to report to the AI Safety Institute before deployment. This includes information about capabilities, safety testing, and risk assessments.
Sector-specific AI rules. Rather than one thorough law, the UK is enableing individual regulators to create AI-specific rules for their sectors. The FCA handles AI in finance, the MHRA handles AI in healthcare, Ofcom handles AI in communications, etc.
AI transparency requirements. Proposals for requiring disclosure when AI is used in consequential decisions — hiring, lending, healthcare, criminal justice. The details are still being worked out, but the direction is clear.
Copyright and AI training. The UK is still grappling with whether AI companies can train on copyrighted material without permission. The previous government proposed a broad exception for AI training; the current government is reconsidering. This is a critical issue for AI companies operating in the UK.
UK vs. EU vs. US
The UK is trying to position itself between the EU’s thorough regulation and the US’s fragmented approach:
More structured than the US. The UK is creating clearer regulatory expectations than the US, where AI governance is scattered across federal agencies and state legislatures.
Less prescriptive than the EU. The UK isn’t adopting the EU AI Act’s detailed risk classification system or its extensive compliance requirements. The approach is more principles-based and gives regulators more flexibility.
The post-Brexit angle. One of the supposed benefits of Brexit was regulatory flexibility — the ability to create rules tailored to the UK rather than following EU directives. AI regulation is a test case for whether that flexibility produces better outcomes.
What It Means for Companies
If you’re building AI in the UK: Expect increasing regulatory requirements, but nothing as burdensome as the EU AI Act (yet). The sector-specific approach means the rules you face depend on your industry.
If you’re deploying AI in the UK: Transparency and accountability requirements are coming. Start documenting your AI systems, their capabilities, and their limitations now.
If you’re operating across UK and EU: You’ll need to comply with both frameworks. The UK’s approach is different enough from the EU’s that you can’t just apply EU compliance and assume you’re covered in the UK.
The Challenges
Regulatory capacity. Asking existing regulators to handle AI on top of their existing responsibilities requires resources and expertise that many don’t have. The FCA, for example, is already stretched thin with fintech regulation.
Coordination. With multiple regulators handling AI in different sectors, there’s a risk of inconsistent approaches and gaps in coverage. The government is trying to coordinate through the AI Safety Institute, but coordination across independent regulators is inherently difficult.
International competitiveness. If UK regulation is too strict, AI companies might choose to base themselves in the US or other less regulated jurisdictions. If it’s too light, the UK might face pressure from trading partners (particularly the EU) to strengthen its approach.
My Take
The UK’s AI regulation approach is sensible in theory — more structured than the US, less burdensome than the EU, adapted to the UK’s specific context. Whether it works in practice depends on execution, and that’s where things get uncertain.
The biggest risk isn’t the regulation itself — it’s the uncertainty. Companies need clear rules to plan and invest. The UK’s evolving approach, while thoughtful, creates a period of uncertainty that could slow AI investment and adoption.
For companies operating in the UK: engage with the regulatory process now. The rules are still being written, and industry input matters. Don’t wait for final regulations to start preparing.
🕒 Last updated: · Originally published: March 12, 2026