The United Kingdom (UK) Parliament’s Joint Committee on Human Rights has called on the British Government to introduce a dedicated artificial intelligence (AI) Bill and establish an independent statutory oversight body, warning that existing laws are inadequate to protect people from the growing human rights risks posed by AI.
The Committee made the recommendations in a 100-page report titled “Human Rights and the Regulation of AI”, its fourth report of the 2026–27 Session, published on September 14, 2026, which provides further arguments for adopting human-rights-based AI governance frameworks before high-risk systems become deeply embedded in public and private institutions.
The report followed an inquiry launched in July 2025 into the implications of AI for human rights and whether the UK’s existing legal and regulatory framework was adequate to address emerging risks.
Although focused on the UK context, it has wider relevance for countries around the world still developing their own AI governance frameworks. The report offers a useful reference point for considering how human rights, transparency, accountability, privacy, non-discrimination, and effective remedies can be embedded in national AI regulation from the outset.
The central message of the report is that AI governance cannot be treated solely as a question of technological innovation or economic policy and that human rights must be built into the design, development, deployment, and oversight of AI systems.
The Committee said AI was developing at a pace that risked leaving governments and existing human rights protections behind, while its growing use by businesses, individuals, and public authorities meant AI systems could affect virtually every aspect of people’s lives.
It warned that the UK’s current legal framework was “patchy and confused”, with significant gaps in protection and redress for people harmed by AI systems.
According to the report, AI could deliver significant benefits, including economic growth and improvements in the protection of human rights, but it also presented novel and serious risks to fundamental rights.
The Committee identified three principal areas of concern, namely equality and non-discrimination, privacy and data protection, and the right to an effective remedy.
It said AI systems were often opaque, making it difficult for individuals to understand why a particular decision or outcome had been generated, adding that people might also be unaware that AI was being used in decisions affecting them.
The Committee warned that the scale and speed of AI deployment meant that when harm occurred, it could affect large numbers of people very quickly.
The risks were considered particularly acute for already marginalised groups, including Black and Minority Ethnic communities.
The report highlighted several examples of AI applications raising human rights concerns, including the use of AI to generate deepfake sexualised images of women and girls, profiling prisoners in ways that disproportionately affected Black and Minority Ethnic people, flagging workers for disciplinary action without reasonable cause, and scanning and recognising people’s faces in public places without their consent.
The Committee also raised concerns about the ability of individuals to challenge AI-generated decisions where they did not know that AI had been used or could not establish how the system had reached its conclusion, saying the problem was compounded by uncertainty over who should be held legally responsible when AI caused harm.
It noted that although numerous laws and regulations could apply to particular AI systems, the overall regulatory landscape remained fragmented as several regulators currently have responsibilities relating to aspects of AI, but there is no single body responsible for coordinating AI regulation across the economy.
This, it said, could leave people affected by AI harms without an effective route to obtain redress.
The Committee was also critical of arrangements that could allow powerful technology companies developing AI systems to shift liability to organisations deploying them, as it argued that responsibility should rest more heavily with those who create and shape AI systems, particularly where they are best placed to identify and prevent potential harms.
It called on the Government to introduce a dedicated AI Bill establishing a comprehensive, risk-based regulatory framework, stressing that the legislation should cover the entire AI lifecycle, from design and development through deployment and subsequent monitoring.
Under the proposed framework, AI systems presenting the greatest risks to human rights would face stronger obligations, while lower-risk applications would be subject to less onerous requirements.
The Committee said some AI applications should be prohibited altogether where they were incompatible with respect for human rights and recommended that the proposed legislation should prohibit uses involving unacceptable risks, including certain applications employing subliminal techniques and other practices capable of causing serious harm.
The proposed AI Bill would also require AI systems posing a high risk of human rights harm to undergo appropriate assessment and controls before deployment.
The Committee recommended obligations covering actors at different stages of the AI lifecycle, including developers and deployers, and called for mandatory human-rights due diligence, with requirements differentiated according to the role and responsibilities of each actor.
It also recommended mandatory transparency, including obligations to tell individuals and communities when AI systems were being used and explain what they were being used for.
The Committee called for stronger safeguards around automated decision-making under the UK General Data Protection Regulation.
It warned that the mere presence of a human “in the loop” should not automatically be regarded as adequate human oversight.
Instead, it said the person reviewing an AI-generated decision should be sufficiently informed about the system and able to exercise genuine independent judgment rather than being improperly influenced by the automated recommendation.
Individuals affected by automated decisions should also be given meaningful opportunities to challenge them.
The Committee recommended establishing an independent AI oversight body, either as a new institution or through the transformation of an existing body, which would coordinate existing regulators, monitor AI systems, investigate breaches, and have enforcement powers.
It would also be responsible for overseeing compliance with the proposed AI Bill, including obligations concerning high-risk AI, human-rights due diligence, and transparency.
The Committee said the body should have powers to intervene where unacceptable risks to human rights were identified and, where necessary, limit or prohibit the deployment of particular AI systems.
It also recommended placing the AI Security Institute (AISI) on a statutory footing with powers to review new and revised powerful AI models and assess their risks.
Developers of powerful AI models should be required to submit new models for evaluation and testing and provide technical specifications needed for meaningful assessments, while the AISI should have a statutory duty to inform the proposed oversight body and the Government about significant risks to human rights identified in AI models and systems.
The Committee rejected the argument that stronger AI regulation necessarily had to undermine technological innovation or economic growth, arguing that regulation could support sustainable innovation by creating clear rules and public confidence while preventing serious human rights harms.
It stressed that the objective should be to promote AI that generates economic and social benefits without sacrificing fundamental rights.
The Committee said the UK should also continue playing a leading role in international discussions on AI governance, given the global nature of AI development and deployment.
The report has wider relevance for countries developing their own AI governance frameworks, including countries in Africa, as the Committee’s concerns about discrimination, privacy, surveillance, automated decision-making, transparency, accountability, and access to effective remedies mirror issues increasingly confronting African countries as governments and private companies deploy AI in different areas, including policing, public administration, employment, financial services, healthcare, education, and media.
The UK Parliament’s inquiry received written evidence from more than 70 organisations and individuals and held multiple oral evidence sessions involving government officials, regulators, technology companies, academics, and civil society organisations.
