Digital Economy Soundbite | AI: the buck stops…where?
20 January 2026
AI technology is changing fast, giving rise to new risks and corresponding questions of liability for physical and economic harm. Lawmakers have been wrestling with the topic of liability for a while, with no clear steer emerging. At an EU level, proposals for an AI Liability Directive have been scrapped. Does English law as it stands provide a ready-made framework? The UK Jurisdiction Taskforce is currently consulting on the point and its January 2026 Legal Statement on Liability for AI Harms examines the state of play.
As the Statement highlights, liability in an AI context raises specific questions:
- AI has no separate legal personality
- Its "black box" nature makes establishing fault very difficult
- As its autonomy increases, human intervention decreases.
Risk can be allocated up and down a supply chain through the usual contractual tools of warranties, indemnities and liability caps. However, where no contract is in place, focus shifts to the established principles of negligence: whether a duty of care exists and, if so, to what standard, and whether there was a failure to meet it; whether that failure caused the loss, and the loss was not too remote.
Of these, causation might prove most difficult in practice. Autonomous decisions by an AI system might not easily map across to the traditional "but-for" test, especially where levels of human oversight drop, and the courts may have to develop their thinking to match this. Otherwise, the principles of vicarious liability and professional negligence apply in the normal way, and organisations will be responsible for chatbots which clearly represent them.
Organisations should take risk mitigation measures that reflect these common law principles. Careful selection of systems, accurate prompting, transparency, AI literacy, and keeping a "human in the loop" will all support a proper standard of care. Nevertheless, disputes will inevitably come to light over the next few years and we can expect some interesting decisions by the courts.
The Taskforce is hosting a discussion on the Statement later this month and we will follow up with further comment.
