Joint committee on Human Rights

Paula Barker Excerpts

Thursday 15th October 2026 Westminster Hall

[Paula Barker (in the Chair)]

We begin with the Select Committee statement. Alex Sobel will speak on the publication of the fourth report of the Joint Committee on Human Rights, “Human Rights and the Regulation of AI”, HC 160, for up to 10 minutes, during which no interventions may be taken. At the conclusion of his statement, I will call Members to put questions on the subject of the statement, and call Alex Sobel to respond to these in turn. Questions should be brief and Members may ask only one question each. I call Alex Sobel, chair of the Joint Committee on Human Rights.

[Alex Sobel (Leeds Central and Headingley) (Lab/Co-op)]

Thank you very much, Mrs Barker. As the newly elected Chair, I am proud to be presenting the significant findings of the fourth report of the Joint Committee on Human Rights, on human rights and the regulation of artificial intelligence. I thank Lord Alton for his exemplary leadership and service as the Chair of our Committee.

All the evidence suggests that the world is at the precipice of the outcomes of a reckless race towards superintelligent AI—AI that can fully replace and outmatch humans at any task. Our report found that no cross-economy regulators, agencies or bodies have systemic powers to prevent AI models or systems from being deployed, even if they pose significant risks.

Regulators lack the power to test and evaluate AI systems before they are publicly released or to prevent their release if they are considered to pose unacceptable risks, except in very limited circumstances. Our existing copyright laws will not stop the onset of superintelligent AI in the UK. Model developers engage the AI Security Institute on a voluntary basis. The AI Security Institute has no statutory power.

The Committee found that AI indeed threatens the human rights of people in the UK and elsewhere. We welcome the Prime Minister accepting one of our key recommendations that the Government should use its G20 presidency to address AI harms. That being said, with the speed and turbulence that frontier developers are hurtling towards unleashing superintelligent AI, we must regulate now—not in a year, not in five years, but now. With the G20 in mind, we should seek a global agreement on standards and regulations next year.

The Committee has concentrated on the key principles identified in the Council of Europe’s framework convention as being fundamental to the protection of human rights— equality and non-discrimination, and privacy and personal data protection. The Committee defined that AI systems are machine systems that perceive, learn and act.

AI is now widely used across a diverse range of purposes, from email spam filters through to powerful tools for scientific discovery. AI systems are being incorporated into more complex AI agents. These are small, specialised pieces of software that can make decisions and operate co-operatively or independently to achieve pre-specified objectives. A fundamental aspect of current AI systems is their capacity to operate with varying degrees of autonomy and adaptiveness. Some AI models operate in ways that are so mathematically complex that their outputs are opaque, even to an informed expert.

The general-purpose AI systems used today are not coded line by line, but use vast amounts of data and are grown, not built. This results in systems that nobody fully understands or can reliably predict.

Let me be clear: AI is the crucial technology of the 21st century. It is vital to effective defence and promotes vast benefits for science, medicine and the UK’s prosperity. However, there are many ways in which AI may adversely affect human rights. In particular, witnesses during our inquiry told us of the potential impact on the treatment and condition of workers, workers’ rights generally and on the environment.

Furthermore, companies are spending billions to make superintelligent AI reality. The chief executives of top AI companies themselves put the chance that it ends all human life at between 10% and 25%. In any other industry, with that level of risk, this development would be ruled out. How have we allowed frontier developers to create the rules in which they govern themselves?

On the first principle of the framework convention—equality and non-discrimination—we heard that intentional abuses committed using AI systems such as deepfakes can have a disproportionate effect on people from minoritised groups, compounding existing structural and systemic racism and sexism that impacts victims’ personal and professional lives. The Safety Not Surveillance coalition told us:

“AI is having a chilling effect on communities of colour in the UK.”

Discrimination can arise when AI systems are deployed, because of these baked-in biases and choices made by those deploying AI systems. We heard explicit examples of AI facial recognition software having racial basis and of AI recruitment tools discriminating against women applicants. AI systems are prone to producing unfairly discriminatory outcomes because of biased training data. That threatens the human rights of people in the UK and elsewhere.

Next, we looked at the second principle of the framework convention, which is that AI systems pose particular risks to privacy and personal data. The Committee found that because of the prevalence of web scraping to collect personal data to train large language models, there is potential for AI systems to be used in highly intrusive ways. That threatens the article 8 right to privacy of people in the UK and is given particular expression in data protection law in relation to personal data. Between January and October 2025, about 3 million people would have had their faces scanned by police facial recognition technology in the UK without their consent.

Human rights principles must inform all aspects of the Government’s policies and practices relating to AI systems, including their approach to the regulation of AI. The Committee recommends that the Government encourage the development of AI use cases that can demonstrably improve human rights.

Although the AI action plan commissioned by the previous Administration brings a welcome focus on innovation and economic opportunity, it fails to pay sufficient attention to the need to secure the protection of human rights from AI-generated interference throughout the supply chain.

The UK’s leading position as the third largest AI market in the world affords opportunities not only for economic growth, but for setting strong standards on the responsible adoption and development of new technologies. The Committee is encouraged by the prominence given to AI by the new Administration and urges the Government to seize the opportunity to regulate AI in a way that puts human rights first. To address the global nature of AI harms and protect human rights, Governments need to collaborate to identify risks and promote consistent standards. The UK can be justifiably proud of its record of leadership on AI. That has been maintained across Administrations from different sides of the debate. An appropriate regulatory regime would not only enhance human rights protections but help to reinforce the UK’s leadership position in AI safety.

The Committee recommends that the Government introduce a new AI Bill, as promised in the 2024 King’s Speech, to place AI regulation on a sound basis and give full effect to the framework convention. A risk-based approach is needed to address AI harms. The Hugging Face incident and others illustrate the significant risks that can arise from the newest and most advanced AI models, and legislation should address those risks on a precautionary basis with a view to avoiding harm, as with the regulation of pharmaceuticals. Low-risk AI systems should be subject to fewer and less demanding legal requirements. To ensure that regulation can keep pace with technological developments, it is preferable for legislation to take the form of broadly framed principles. The AI Bill that we recommend should provide for a regulatory regime that would classify different risk levels and mandate proportionately more demanding obligations for higher-risk AI systems and models.

Some uses of AI should be prohibited because they are incompatible with human rights. I believe that to be the case when it comes to superintelligent AI. Once developers reach superintelligent AI, it will be too late to slow anything down in time for us to regulate. The development of the most advanced AI systems happens on a spectrum, so where we draw the line matters. That line must be able to recognise the capabilities of superintelligent AI. The clearest example of such a capability is recursive self-improvement—the full automation of AI development, with no humans in the loop. That is the point at which we lose control, as development towards superintelligent AI would proceed too fast for humans to intervene and remain in control.

Frontier developers are heeding standards, rather than statute, for their own AI models. These are the same developers who are deliberately training their models to hack and deceive, as we heard this week during the Business, Innovation, Science and Trade Committee inquiry session, which I guested at as JCHR Chair. We heard from ex-employees of frontier developers OpenAI, Google DeepMind and Anthropic. We heard about non-existent safety cultures in these organisations. Some whistleblowers put the chance of AI destroying humanity at 50%—a coin flip as to whether we survive. This is not a human rights model that we should ever be working with. These frontier developers believe that they will not get stopped by Government in their race to superintelligence. Let us put a stop to that and regulate now.

In the words of David Krueger, previously at the UK AI Security Institute,

“My dream is that in 10 years people will look back and say: ‘Remember when there were all these people building AI that could’ve killed everybody? That was CRAZY! What were we thinking!?’”