UK AI Guidance for Regulators – February 2024

The U.K. Department for Science, Innovation and Technology (DSIT) last week published its new paper “Implementing the UK’s AI Regulatory Principles: Initial Guidance for Regulators.”

The U.K.’s approach to AI regulation relies on existing regulators to identify and implement whatever changes and guidance they think are required, within their own regulatory remit. The new DSIT guidance for regulators is voluntary. It sets out suggestions for how regulators could consider each of the five regulatory principles for AI: (1) safety, security and robustness; (2) appropriate transparency and explainability; (3) fairness; (4) accountability and governance; and (5) contestability and redress.

This is all well and good, but innovators in regulated industries want to adopt tech now. They want to know: 1) will our regulator(s) be making any changes to regulation? And 2) how can we ensure that our use of AI will be compliant?

Where AI-innovators might find the DSIT’s initial guidance helpful, is that: 

1. It sets out at Annex 1 all existing guidance from U.K. regulators. The ICO’s guidance in particular will have broad applicability (see “Guidance on AI and data protection" and "Explaining decisions made with AI,” co-authored with the Alan Turing Institute).
2. When discussing each of the five principles, it sets out the specifically relevant guidance for each principle from U.K. and other regulators, as well as relevant technical standards (e.g., ISO standards).

The guidance also encourages U.K. regulators to collaborate more with each other and to improve transparency on what they are doing regarding AI regulation. Transparency and communication are important even where a regulator adopts a “technology agnostic” approach (as the Financial Conduct Authority has done). Even if the regulations themselves are not going to change, firms still need regulators to guide them on how to ensure their AI use is compliant. 

The initial guidance from the DSIT is phase one of its planned guidance, with phases two and three intended to provide more detail following input and collaboration from regulators and other stakeholders.

For AI-regulation in the EU, see our colleague Morgan Jones’ article here.