The life of the law has not been logic: it has been experience.
Oliver Wendell Holmes Jr, The Common Law (1881)
Much of what has been sold as “AI governance” in the UK is old law in new packaging: the obligations that bite on the use of AI are the ones organisations already have, in data protection, equality, negligence and contract, and, for governing bodies and professional bodies, in their own rules. That is, on reflection, rather good news, since it means the answer to AI lies mostly in doing existing compliance properly, rather than in building a parallel bureaucracy alongside it.
Why has “AI governance” struggled to find its footing?
Because it arrived ahead of the law it claimed to serve. The UK has so far chosen to regulate AI through its existing regulators and existing legal principles rather than through a single AI statute, and yet a whole market of frameworks, registers, certifications and platforms has grown up as though such a statute were already in force. For most organisations the result has been a stack of documents that nobody owns, describing obligations that are either already covered elsewhere or do not, in fact, exist, and a fairly widespread, and not unreasonable, sense that the subject is more noise than substance.
What law actually applies to the use of AI?
The same law that applied before. The ICO’s guidance on AI and data protection holds that UK GDPR applies in full to AI systems, whether they were built in-house or bought from a vendor, so the questions are the familiar ones of lawful basis, transparency, accuracy and security. Equality law applies to the outcomes AI produces, and the burden sits with the organisation using the tool rather than the developer; “the algorithm did it” is not a defence. The UK Jurisdiction Taskforce’s Legal Statement on Liability for AI Harms (July 2026) confirms that English law already holds those who use AI negligently liable for the harm that results, on ordinary principles and without waiting for anything new. Copyright, contract and confidentiality all bear on what goes into a tool and who owns what comes out of it.
For membership bodies there is a further layer, and it is the one most often overlooked: the rules themselves. A conduct code, a disciplinary procedure or an assessment regulation either reaches AI-assisted conduct or it does not, and if it does not, the gap will be found in a live case rather than in a policy review.
Where does that leave organisations?
In a better place than the marketing suggests. None of this is a reason to avoid AI, and nor is it a reason to build a separate governance apparatus for it. It is a reason to make sure that the policies, contracts and, for membership bodies, the rules an organisation already has are written clearly enough to apply when AI is involved, and to check them in the ordinary course of a rules review rather than as a special project.
We don’t ban axes. We train woodcutters, and we make sure the rules of the forest are clear about who may cut what. The same logic applies to AI, and it is, at bottom, the ordinary business of compliance. Read more about our Regulatory Compliance advisory services.