The UK’ government’s “deregulatory” approach to artificial intelligence (AI) will fail to deal with the technology’s highly scalable harms, and could lead to further public disenfranchisement, the Parliamentary inquiry on artificial intelligence (AI) and human rights has heard.
Launched in July 2025, the inquiry was set up to examine how human rights can be protected in “the age of artificial intelligence”, with a particular focus on privacy and data usage, discrimination and bias, and creating effective remedies for situations where AI systems cause violations.
Speaking during the inquiry’s second evidence session on 29 October 2025, expert witnesses told Parliament’s Joint Committee on Human Rights that, as it stands, the UK’s “uncritical and deregulatory” approach to AI will fail to deal with the clear human rights harms presented by the technology.
This includes harms related to surveillance and automated decision-making, which can variously impact both collective and individual rights to privacy, non-discrimination, and freedom of assembly; especially given the speed and scale at which the technology operates.
“AI is regulated in the UK, but only incidentally and not well … we’re looking at a system that has big gaps in [regulatory] coverage,” said Michael Birtwistle, the Ada Lovelace Institute’s associate director of law and policy.
“We don’t have regulators in many high-impact contexts that AI is being used in – employment, recruitment, large chunks of the public sector like benefits and tax administration,” he said. “Our education regulators don’t explicitly have scope around tech use in schools, for example.”
Birtwistle added that while the AI opportunities action plan published by the government in January 2025 outlines “significant ambitions to grow AI adoption”, it contains little on what actions can be taken to mitigate AI risks, “and no mention of human rights”.
Deepening disenfranchisement and discrimination
The experts present also warned that the government’s current approach – which they said favours economic growth and the commercial interests of industry above all else – could further deepen public disenfranchisement if it fails to protect ordinary people’s rights, and makes them feel like technology is being imposed on them from above.
Witnesses also spoke about the risk of AI exacerbating many existing issues, particularly around discrimination in society, by automating processes in ways that project historical inequalities or injustices into the future
On growing disenfranchisement, for example, Birtwistle highlighted a range polls looking at public attitudes towards AI, which all show that the public want to prioritise fairness, safety and the positive social impacts of the technology, over and beyond economic benefits, speed of innovation, or concerns around international competition.
“There is a strong sense of public disenfranchisement around AI,” he said. “Many people don’t feel like they have a say in what the government does, and it’s coupled with high levels of concern about whether the government will prioritise relationships with Big Tech, for example.”
Silkie Carlo, director at privacy campaign group Big Brother Watch – who also highlighted the government’s “very optimistic and commercial-focused outlook” on AI – noted that so far in this Parliament, the Data Use and Access Act (DUAA) has “decimated” people’s protections against automated decision-making, by essentially removing the “assumed prohibition” and liberalising it for all purposes.
She added that the government should instead place more weight on the inherent risks of AI, which can produce harm or injury at speed and scale in a way that other technologies are simply not capable of.
Facial recognition
Giving the example of AI-driven facial recognition being deployed by UK police, Carlo said that while it is true the technology is getting more accurate and making fewer misidentifications, the rate and scale of the data processing taking place means even if only a very small percentage of people are negatively affected, that is thousands of people impacted in real terms.
“There is very little accountability and transparency with facial recognition, but other types of AI as well,” she said. “If you are flagged by many of these systems, you may not know why, and I think that contradicts very basic British values. It also means that there’s a chilling effect, because if you feel that surveillance systems or other AI systems used by government departments could be watching you, judging you, flagging you, making decisions about you, and you don’t know why or you can’t challenge it, I do think that cuts through our very long-standing values about fairness, and what we expect from public authorities.”
Carlo added that there is real potential for AI-enabled mass surveillance to “spiral out of control”, and that a system built for one purpose could easily be deployed for another “in the blink of an eye”.
On the potential for AI to exacerbate social problems and discriminatory outcomes, tèmítópé lasade-anderson, executive director at anti-discrimination charity Glitch, highlighted how existing police datasets tend to massively over-represent people from lower socioeconomic backgrounds or particular ethnic minorities.
Giving the example of how postcodes are used as proxies for different demographics of people, she added that lower socioeconomic areas tend to have more engagement with police.
This is then reflected in the data collected, which in turn prompts more police engagement with that area, consolidating the notion there are issues as the data collected and police activity form a negative feedback loop.
“When you apply that sort of database into an automated tool, it is going to likely suggest that these people from this area need to be surveilled,” she said, adding that “debiasing” datasets is “simply not good enough” because it will not deal with the underlying social issues that mean people are being discriminated against in the first place. “There needs to be some sort of red lines on where AI is not involved.”
Public engagement and effective redress
Witnesses also discussed in more detail what would be needed for effective AI regulation in the UK, which included the creation of sector-specific rules that deal with the nuances of the technology’s highly contextual use, and structuring these regulations so that they capture every part of a given system’s life cycle.
They also said an emphasis was needed on “co-creation” to build trust with the public and avoid the imposition of technology from “on high” by Silicon Valley or faceless government bureaucracies.
Speaking during the committee’s first session on 2 July 2025, the Alan Turing Institute’s director of ethics and responsible innovation research, David Leslie, also previously said there is a pressing need to “shape innovation ecosystems so there is more, rather than less, empowerment of members of the public”.
“The public need to be involved in these decisions about using, designing and deploying systems and data collection so that there can be structural public determinations of what is and is not acceptable,” he said.
However, across both sessions, witnesses were clear that all of this would need to be underpinned by creating powerful mechanisms for redress, which would allow people to effectively challenge AI systems and protect their rights.
“The [current] mechanisms for actually exercising your data rights and your human rights … will not function for most people,” said Birtwistle, who also highlighted how the problem of ineffective redress is being magnified by the government’s DUAA, which has broadened the set of conditions where automated decision-making is permissible in the UK. “It’s very difficult to get redress. In practice, it’s very expensive, and you’re unlikely to succeed.”
Also speaking during the committee’s first evidence session, AWO solicitor and legal director Ravi Naik agreed that while regulatory elements like accountability and transparency are important, and would need to apply to the entire “chain of actors involved”, the key factor is redress; something that is already hard to come by under the UK’s current human rights framework and cost-prohibitive legal system.
On the human rights framework, Naik noted that it currently pertains to public actors, while AI systems are often created and controlled by private technology companies that “tend to keep a lock on” their decision-making and development processes.
“I would very strongly urge this committee to consider the ability of regulators to look at how systems are made and why decisions are being made,” he said.
“Many systems are said to be proprietary and covered by trade secrets legislation; however, that is often a smokescreen because the models are publicly available through open source, for instance, through a platform called Hugging Face. Regulators may have the power to investigate, but they also need to have the skills to look at technology. It cannot be incumbent on individuals to take such action.”
Naik also commented on the cost of bringing action in the UK, particularly against private actors, which is prohibitive for most ordinary people. “We are talking seven-figure sums to bring action against a private actor just to uphold your rights, let alone seek damages,” he said. “That has to be a key barrier to accountability in this space.
Naik suggested to the committee that legal aid should be extended to people so they can take action against private actors, and not just the government. “If a government actor has committed a human rights wrong, most people will get legal aid to prosecute their case,” he said. “The legal aid system exists to make government accountable; that is the very purpose of it, and it is a really powerful tool to ensure accountability.
“The reason legal aid is extended to government actors was a question of power, but I would question who has power now,” said Naik. “In the digital space, a lot of power resides in private hands. So, when we are thinking about how to try to redress a wrong, those are some key concepts and hurdles that are before us.”
