Source: Allen & Overy (A&O Shearman insight)

The UK Jurisdiction Taskforce (UKJT) presents a legal statement asserting that English common law can address non‑deliberate harm caused by AI without new bespoke regulation, with courts expected to adapt existing doctrines incrementally by analogy. Contractual allocation of risk is identified as the primary route for distributing liability across the AI supply chain; where contract does not govern, the UKJT suggests negligence will most likely provide the relevant framework. Duties and standards of care are described as highly fact‑specific and may reach foundation model developers, application developers and deployers, and professionals using AI tools, with potentially surprising positions such as software updates supporting an assumed duty of care, professionals being negligent for failing to use AI, and developers possibly liable for creating a “source of danger.”

The UKJT adopts a concise definition of AI as “a technology that is autonomous” and highlights AI‑specific difficulties: unpredictable input–output relationships, opacity of reasoning, and limited user control. Practical uncertainties include complex supply chains, potential scale and diffusion of harm, rapid system operation, and human–AI interaction. Evidential and causation issues are expected to be central, making record‑keeping, due diligence, disclaimers, transparency and human oversight important safeguards. The statement notes that strict liability routes are narrow (the Consumer Protection Act 1987 typically applies only where AI is part of a physical product), records the Law Commission’s July 31, 2025 review intention on “pure software,” discusses vicarious and non‑delegable duties (with an NHS Trust example and potential recourse under the Civil Liability (Contribution) Act 1978), and cautions that many outcomes will be highly fact‑sensitive.