The harms are here: why Burnham should put human rights at the heart of the Government’s AI Strategy

The Hugging Face hack and the subsequent string of frontier AI lab researchers raising the alarm about existential risk has launched AI’s dangers back into the headlines.

Whilst these hype and doom narratives are baffling to anyone not deep in Silicon Valley’s international networks, these risks deserve our attention and scrutiny. Where they prove dangerous and imminent, like in cyber security, the Government should act. However, we cannot let the panic about future harms divert our attention from the harms that have already arrived.

You fall in love and you decide to get married, but your application for a marriage licence is rejected because your partner is Greek and a DWP algorithm flagged you as a potential scammer.¹ You study in the library every weekend for a year, determined to make your A-Level grades to get into your dream university, but because no one in your school had ever gotten an A* before, an algorithm pulled you down and you missed your offer.² You are pulled out of your home in the middle of the night for a committing a crime in a city you’ve never been to because a biased algorithm matched your face incorrectly.³ These are all examples of the life-changing impacts of government AI systems already going wrong.

Liberty has been watching as AI has become the plaster to stick on many of the Government’s problems. With ailing public services and little money to fix them, AI and automation offer shiny solutions that seem to answer to successive governments’ complaints about the state (too slow with too much bureaucracy). Starmer hailed AI as the ‘defining opportunity of our generation’⁴ and a development with ‘revolutionary’ potential.⁵ However, amidst the excitement, the Government has consistently undervalued the risks that ill-thought-out AI adoption automates into our state.

We are particularly concerned about the accelerating use of AI in policing and criminal justice. We have already seen police officers using AI systems to try to influence criminal justice outcomes.⁶ Even routine uses of AI like redaction, transcription, and report drafting can raise issues of bias and accuracy, with potentially serious consequences for the evidence chain. Higher risk systems like facial recognition and predictive policing are already impacting people’s lives: Alvi Choudhury was arrested after facial recognition incorrectly matched him to a suspect in a city he had never been to.⁷ If the Government continues to roll out AI without proper safeguards, more people will be misidentified or flagged by racially biased algorithms, and more people will have their access to public services and their criminal justice outcomes shaped by AI scores they never see or understand.

The UK has long been the one of most surveilled democracies in the world. However, police forces and intelligence agencies have lacked the manpower to use all their camera networks and data to their full extent. Now, with AI systems able to scrape these massive data sources and make them useful, it will hand huge power to governments. We need only look to the US to see how this can be used to target migrants, activists, and journalists.

It is tempting, in the face of technological innovation and a hype cycle driven by those innovators, to think of this moment as unprecedented. Of course, AI creates new, unique challenges for tech companies, policymakers, and for society. But this is not the first time we have grappled with a new technology.

Human rights frameworks emerged out of the destruction wrought by a technological revolution. Railways, census data, communication systems, the administrative processes of the factory – the machinery of the modern, industrialised state – gave Hitler’s genocidal regime unprecedented power over its people. Whilst the concepts that underpinned human rights frameworks had been developing for centuries, it was in the wake of the Holocaust that they were adopted and codified en masse and the world agreed: ‘All human beings are born free and equal in dignity and rights.’⁸ The scale of the war’s destruction crystallised the importance of each human life, and the rights that make a life worth living.

Technological innovation is once again increasing state (and non-state) actors’ power over our lives. We should not wait for tragedy to act to constrain this power again, and human rights is the toolkit we built to do so. Human rights also provide something that AI safety, transparency, and ethics cannot: a universal set of values and principles that we have collectively agreed are important. They are globally salient; 80% of UN member states have ratified four or more of the core international human rights treaties.⁹ This global recognition makes human rights the best foundation for aligning AI with values that resonate across cultures.

Human rights seize the discussion about AI and pull it away from machines and towards the human. They are a way of inserting humanity and our needs back into these systems and re-asserting what is useful and important to our lives. It is not inevitable that innovation should erode our freedoms and our dignity.

As Burnham begins his first Parliament as Prime Minister, he should take the opportunity to turn a new leaf and reset the Government’s approach to AI. The Labour Government must take a human first approach to technology policy. Putting human rights at the heart of his approach would be the first step towards building better tech for all Britons.

1 Kiran Stacey, ‘UK officials use AI to decide on issues from benefits to marriage licenses’, Guardian, 23 October 2023.

2 Pamela Duncan, Niamh McIntyre, Rhi Storer and Cath Levett,  ‘Who won and who lost: when A-levels meet the algorithm’, Guardian, 13 August 2020.

3 Robert Booth and Mark Wilding, ‘Facial recognition error prompts police to arrest Asian man for burglary 100 miles away’, Guardian, 25 February 2026.

4 Rt Hon Sir Keir Starmer MP, ‘PM speech on AI Opportunities Action Plan’, Prime Minister’s Office, 10 Downing Street, 13 January 2025.

5 Ibid.

6 Rebecca Sherdley, ‘Senior detective under investigation over use of AI’, BBC News, 10 August 2026.

7 Robert Booth and Mark Wilding, ‘Facial recognition error prompts police to arrest Asian man for burglary 100 miles away’, Guardian, 25 February 2026.

8 Universal Declaration of Human Rights, Article 1

9 UN Treaty Body Database, United Nations Human Rights Treaty Bodies website, accessed 02/09/26.