Time waits for no one: an AI Bill is imminent

The AI Bill aims to regulate artificial intelligence technologies given the rapid advancement of AI over the last year.

Governments around the world have recognised the need for legislation to govern AI development and deployment. The EU AI Act has been published at the end of last week in the Official Journal, meaning the Act will come into force at the beginning of August (with the various chapters having different implementation dates over 2025-2027).

The UK’s pro-innovation approach to date has done nothing more than confirm that guidance won’t be given to AI service providers, developers and users. The new Government are opposed to this approach, and it's been reported in the Financial Times following a conversation with the leader of the House of Commons, Lucy Powell, that they've been working on what “are fully considered, worked-up bills that we know we can get through in this parliamentary session.”

What will the UK AI Bill include, and what could it mean for AI development in the UK?

At present we can only really defer to the Labour Manifesto 2024, which states that:

• A Regulatory Innovation Office will be created which brings together existing functions across government to help regulators update regulation, speed up approval timelines, and co-ordinate issues that span existing boundaries.
• The safe development and use of AI models by introducing binding regulation on the handful of companies developing the most powerful AI models, will be a key focus.
• There will be a ban on the creation of sexually explicit deepfakes.

A logical conclusion is that the Bill will cover ethical AI considerations such as fairness, accountability and transparency, as well as data privacy and bias mitigation. There are likely to be compliance requirements to implement audits, explainability and accountability mechanisms. Liability is likely to be of particular importance to ensure developers are held to account. Further, it’s also possible that AI regulatory sandboxes will be created to allow businesses to test innovative propositions in the market with real consumers.

The Bill could have extra-territorial effect, much like the EU AI Act (and General Data Protection Regulation 2016, Privacy and Electronic Communications Regulations 2003 and Data Protection Act 2018 in the UK), meaning that it could extend to organisations that place, make available, or put into service AI products in the UK. Indeed, there are benefits to the Bill aligning with the precedent set out in the EU AI Act not least from an international harmonisation perspective. 

The Bill must encourage responsible development without stifling creativity. In any event, it can bring several benefits giving UK companies adhering to ethical AI practices a competitive edge.

How quickly could it land?

Online sources suggest that the typical period is 153 days (ie, just before Christmas – what a present that would be) for Parliament to pass a bill in the UK (excluding the Royal Assent period and Commencement). Given the importance of the Bill, it is expected that it will go through the usual layers of scrutiny:

• First reading: The formal presentation of the Bill, without any debate.
• Second reading: The House debates the general principles of the Bill. No amendments are made at this stage.
• Committee stage: Detailed examination of the Bill by a committee. Amendments (proposed changes) can be made here.
• Report stage: Further consideration of the Bill, including additional amendments if needed.
• Third reading: The House decides whether to agree with the Bill in its final form.
• Passing to the other House: If the bill started in the Commons, it now goes to the Lords (and vice versa).
• Agreement on Amendments: The first House considers any amendments made by the second House.
• Royal Assent: Once both Houses agree, the Bill receives Royal Assent and becomes law.

What can organisations do in the meantime?

1. Monitor: Stay informed about the progress of the Bill.
2. Risk management: Look at your AI governance practices – to include record keeping.
3. Internal audit: Review existing AI contracts, processes and AI systems.
4. Legal review: Engage legal experts with expertise and seek advice on interpreting the Bill’s requirements and implications tailored to your organisation.
5. Education: Educate employees about the impact of the Bill on their role, including providing them with practical training on ensuring compliance.

How can we help?

As we’ve said before in our previous article, “A whistle-stop guide to managing disputes risk from the development of AI”, assemble a legal dream team that understands the nuances of AI, data protection and technology law to help you bridge the gap between the tech and legal worlds.