Background
Parliament’s Joint Committee on Human Rights (JCHR) has waded into the AI debate with a blunt verdict: the UK’s current approach to regulating artificial intelligence simply isn’t fit for purpose. In a 100-page report published this week, the cross-party group of MPs and peers is calling for a dedicated AI Bill and a new statutory regulator with real teeth to police the technology’s most dangerous uses.
What’s Actually Being Proposed
The committee wants a risk-based regulatory regime covering the entire AI supply chain, developers as well as the businesses deploying their tools. Lower-risk uses would face light-touch obligations, but higher-risk practices, including AI used to predict someone’s likelihood of committing a crime, infer emotions at work, or carry out real-time biometric surveillance, would face far tougher scrutiny, with some uses banned outright pending public consultation. Committee chair Alex Sobel wants a single regulator “with the teeth to ensure enforcement,” backed by mandatory transparency rules so people know when AI has played a part in decisions affecting them.
The Harms Driving the Push
The report isn’t working in the abstract. It cites AI-generated sexualised images of women and girls, facial recognition scans carried out without consent, and cases where automated decision-making left people with no meaningful way to challenge the outcome. Sobel was direct about the urgency: “We need these protections in now, it cannot wait until [these] human rights risks become reality.”
“Human in the Loop” Isn’t Good Enough
One of the report’s sharper conclusions is that simply having a person “review” an AI decision doesn’t count as meaningful oversight. A human check only means something if the person involved is genuinely “sufficiently informed and independent,” not just rubber-stamping a decision the machine already made.
Setting the Political Scene
The report lands against a backdrop of real unease inside the AI industry itself. Anthropic’s Dario Amodei said this weekend that development has been “advancing drastically faster” since the summer, warning that within six to twelve months AI could be capable of coordinating a swarm powerful enough to threaten the internet itself. A departing Anthropic researcher separately told the BBC that staff building these systems were “genuinely frightened” for humanity’s future, though an Anthropic spokesperson pushed back, saying the company builds “models with some of the strongest safeguards in the industry.”
On the government side, education minister Georgia Gould insisted ministers were “responding to those risks,” pointing to action on deepfakes and cyber security legislation, even though the dedicated AI Bill promised in Labour’s 2024 manifesto and the 2024 King’s Speech still hasn’t materialised. Since arriving in No 10 in July, Prime Minister Andy Burnham has scrapped the Department for Science, Innovation and Technology altogether, splitting its responsibilities elsewhere in government while ensuring AI Minister Kanishka Narayan retains a seat at cabinet.
Technology Tuesday Takeaway
Two years on from that manifesto promise, the gap between what Labour said it would do on AI and what’s actually reached the statute book is becoming harder to ignore. MPs on all sides now want developers, not just the businesses using their tools, held accountable for the harms AI can cause. Whether that finally becomes law, or waits for the next headline-grabbing AI failure to force the government’s hand, is the real question this report leaves hanging.
Written by Phoebe De Oliveira Simões
15 September 2026