Mind the AI gap: are human rights adequately protected?

A report from Parliament’s Joint Committee on Human Rights, published on 14 September 2026, calls for changes to the UK’s legal and regulatory framework to address the human rights risks posed by artificial intelligence.

The Committee’s report, Human Rights and the Regulation of AI, concludes that the current framework is fragmented, difficult to navigate and places too much responsibility on those deploying AI systems, rather than developers and providers further up the supply chain. In particular, it warns that existing laws do not adequately account for where risks may originate during the development of AI systems.

Transparency and redress are also key concerns. The Committee highlights that individuals may not know when AI has been used in decisions affecting them, making it difficult to challenge those decisions or seek an effective remedy. It also points to potential risks across a range of rights, including the right to a fair trial where AI is used in court proceedings.

The Committee is calling for a dedicated AI Bill covering the AI lifecycle and supply chain. Its recommendations include:

• a single, independent statutory AI oversight body with meaningful enforcement powers;
• mandatory transparency requirements across the AI lifecycle;
• due diligence obligations on developers, providers and deployers, calibrated according to their role and the level of risk;
• prior approval for AI systems posing a high risk of human rights harm; and
• consideration of outright prohibitions on certain uses of AI that are incompatible with human rights.

CMS comment

The UK’s current approach to AI regulation remains primarily sector-led and is implemented through existing legislation, regulatory guidance and voluntary measures, although the government has continued to develop its wider AI regulatory framework. The Committee Report nevertheless signals increased political concern that this approach may not provide a sufficiently coherent or enforceable framework for managing human rights risks in the age of AI.

If adopted, the proposed measures could materially increase obligations across the AI supply chain. Developers and providers could face greater expectations around design-stage risk assessment, documentation, testing, transparency and information-sharing, while deployers may need to demonstrate that systems have been appropriately assessed, monitored and governed in practice. A central oversight body with enforcement powers could also create a more consistent supervisory regime and increase the risk of regulatory intervention where organisations cannot evidence effective controls.