Westminster’s AI Silence
427 MPs haven’t uttered ‘AI’ once this parliament
Since the new Parliament commenced on 9 July 2024, only 223 of the 650 MPs[1] have mentioned artificial intelligence even once in the main chamber, and just 48 have done so more than three times.
That silence matters. AI systems are already spilling out of the lab and into everyday life, yet Westminster still seems to treat the subject as a sideshow.
Why should MPs be talking about AI?
It is difficult to walk across the tech landscape without bumping into another billionaire proclaiming that super-intelligent, general AI is just around the corner. Scepticism is understandable, but hyping the future can boost a CEO’s influence and their company’s share price. However, big claims do not come only from Silicon Valley founders. The non-profit AI Futures Project recently released its AI 2027 scenario[2], forecasting that automated R&D could produce AIs far beyond human capability by the end of this decade, raising the risk of conflict and widespread human disempowerment.
You might still file that under science fiction. Even so, today’s data shows AI deserves a place at the top of the political agenda. Anthropic finds that AI tools are now used for over a quarter of tasks in over one-third of occupations[3], albeit mainly for augmentation rather than replacement at this stage. In customer-support pilots, productivity jumps of 15 per cent have already been recorded[4]. What those snapshots miss is the speed of improvement. Gemini 2.5-Pro, OpenAI o3, and Claude 4 all outperform the models available only months ago, and even more capable systems are likely in the lab pipeline.
A 15 percent productivity gain here or a 33 percent job-task replacement there might not sound transformative, but for the wider economy, they are seismic. The UK’s annual productivity growth limps along below 1 per cent. Shifts in AI’s scale will produce winners and losers, upheaval and volatility. Effects may already be being seen in the job market. National data from the United States shows recent graduate employment prospects are as bad as they have ever been relative to general employment[5]; anecdotal evidence is building up[6][7], pinning at least some of the blame on AI.
Even if progress in AI froze tomorrow, the penetration of existing systems alone would still bring jolts that Parliament cannot afford to ignore.
Does the House of Commons still matter?
Some scholars argue that Parliament is drifting toward rubber-stamp status. Others insist the real action now takes place between Downing Street and California’s boardrooms, or in bilateral summits between leaders. Yet when the political weather turns, the Commons still bites. Forcing U-turns on Syrian air-strikes and Brexit, and, more routinely, nudging ministers to rewrite clauses or drop proposals the moment they sense the numbers are against them. The Commons remains the venue where competing visions are thrashed out, amendments are forced, and ministers are made to explain themselves on record. It remains the elected bridge between the public and policy. That is exactly where a technology with deep technical, economic, and moral consequences ought to be interrogated.
What MPs have said so far
Between 9 July 2024 and 31 May 2025, every Commons sitting was scraped for references to “AI” or “artificial intelligence”. Each mention was logged to the speaker and run through a basic sentiment model to gauge tone.
223 MPs mentioned AI at least once, and 48 more than three times. For context, over the same period, a small but divisive issue, the Winter Fuel Payment, drew mentions from 210 MPs at least once and 79 MPs on three or more occasions[8]. When a once-a-year benefit almost matches a potentially epoch-shaping technology, something is off.
Unsurprisingly, Peter Kyle, Secretary of State for Science, Innovation and Technology, dominates the table for AI mentions. Chris Bryant appears frequently in his overlapping junior-ministerial briefs. Most of the other heavy contributors fall into two camps: ministers (Pat McFadden, Shabana Mahmood, Feryal Clark) and select committee or shadow-team leads (Chi Onwurah, Victoria Collins, Caroline Dinenage). Only two backbenchers break into the top ten. The SNP’s Pete Wishart and Conservative John Whittingdale have both focused on copyright and creator rights in the AI economy. While that topic deserves attention, it is striking how few colleagues have adopted to a significant degree any of the other looming questions: election integrity, national-security risk, and the impact on labour markets.
Sporadic AI Scrutiny is Falling Short
Away from sporadic questions, most discussion in the house has been related to the Data (Use and Access) Bill. Debates on this bill have been held both in the main chamber and Westminster Hall alongside significant engagement from the House of Lords. This bill, initially introduced in October 2024, seeks to modernise the UK’s data protection framework. Amongst its main objectives are to enable more data sharing within the government. Some of its broader aspects have seen resistance from the creative industries concerned about an ‘opt-out’ vs ‘opt-in’ for sharing copyrighted work.
Beyond this bill, meaningful parliamentary engagement has been minimal. Only the AI Opportunities Action Plan has generated additional noteworthy debate. Across select committees, including Science, Innovation and Technology, Defence Sub-committee, and Treasury, AI-focused workstreams exist but remain fragmented and lack prominence or a coherent, cross-cutting approach.
This contrasts sharply with the United States, where engagement has surged to unprecedented levels. Over 1,000 AI-related bills[9] have been introduced at the state and federal levels. Few will reach the statute book, but the sheer pipeline sets a sharp contrast with Westminster. Beyond bill drafting, in 2023 alone, House and Senate committees conducted over 30[10] hearings focused on AI issues, rising to more than 50[11] in 2024. Congress is even now moving to establish a moratorium on AI legislation to avoid a messy patchwork of differential state legislation[12]. Whether local or national, policymakers are clearly engaged. While Silicon Valley’s proximity and the clout of its tech giants may explain part of the gap, it’s hard to overstate the difference in legislative interest and discussion.
Sentiment scores don’t change the headline. MPs aren’t talking about AI enough, regardless of their views. But tracking the tenor of the few speeches we do have adds useful colour to the scant debate happening in the Commons.
How AI sounds in the chamber
Labour’s tone is the most upbeat, skewed by Peter Kyle’s strong promotion of AI-driven growth and the potential for public sector efficiency gains (average sentiment c.+0.7). Conservative and Liberal Democrat contributions are more measured. A deeper dive into sentiment and issue clusters. Who worries about jobs, who about geopolitics, who about copyright? Will follow in a separate piece. It is worth noting, however, that sentiment analysis using general models is coarse-grained and, despite some promising studies[13], reading too much into results is cautioned against.
Westminster’s AI vacuum
Even allowing for front-bench dominance, the numbers are hard to ignore: 427 MPs have not said “AI” in the main chamber at all this Parliament, and fewer than fifty can be called regular contributors. AI research moves on a month-by-month cycle, sometimes week-by-week. Parliamentary scrutiny still moves on an archaic sitting-day timetable. Unless that gap closes, the UK risks waking up to labour-market shocks, security dilemmas and public-service upheaval it has barely debated.
Time to raise the volume
AI is not a niche research topic. It is already eating into workloads, rewriting military doctrines and nudging global power balances. The House of Commons will have to grapple with those realities soon; better by deliberate debate now than by rushed legislation after a shock.
But debate alone isn’t enough; Westminster needs discussion and concrete follow-through. Three practical steps:
1. Dedicate parliamentary time for substantive, open discussion on AI’s societal and economic impacts, exploring broad principles and strategies for regulation.
2. Create a joint select committee inquiry across multiple policy areas (Science and Technology, Defence, Treasury, and Home Affairs), ensuring detailed, cohesive, and high-profile cross-party scrutiny on AI policy implications.
3. Commit to comprehensive AI training for MPs. While it is unclear the extent to which the lack of discussion is due to a lack of knowledge. AI is an immensely complex and multifaceted field. Education is required to legislate and scrutinise this rapidly evolving technology effectively.
If Westminster cannot supply clear answers, someone else will. Perhaps corporate lobbyists, perhaps foreign regulators. The choice facing Parliament is whether to lead the conversation or chase it from behind.
More on the method
- Data source: official Commons Hansard transcripts, captured daily.
- Keywords: “AI” and “artificial intelligence”. Other terms (LLM, machine learning, etc) were excluded for clarity. Checks of the major debates show use of other terms is minimal — 12th of Feb 2025 was the largest single day for AI/artificial intelligence mentions with 69 — zero mentions of LLM or ChatGPT, one mention of machine learning.
- Counting rule: multiple hits inside one speech passage count as a single mention; long ministerial statements with interventions can thus inflate totals.
- Westminster Hall debates are not included in the current analysis.
- Speaker data: Dynamic Hansard-supplied IDs, with junior-minister titles added manually where missing.
- Sentiment: whole-passage scores via a single prompt to ChatGPT 4.1-mini-2025–04–14. The model is crude; it treats complex policy nuance as neutral.
Limitations remain. Sentiment is coarse, certain debates may dominate totals, and Commons debates are only one slice of Westminster-wide activity.
[1] Speaker and three deputies do not speak on political matters, 7 Sinn Fein MPs do not take their seats, this is not factored into headline numbers
[3] https://www.anthropic.com/news/the-anthropic-economic-index
[4] Erik Brynjolfsson, Danielle Li, Lindsey Raymond, Generative AI at Work, The Quarterly Journal of Economics, Volume 140, Issue 2, May 2025, Pages 889–942
[5] Federal Reserve Bank of New York, The Labor Market for Recent College Graduates, https://nyfed.org/collegelabor.
[6] https://www.ft.com/content/8e730692-fd9c-45b1-84dc-7ea16429c5c6
[7] https://www.zdnet.com/article/is-ai-making-it-harder-for-new-college-grads-to-get-hired-in-tech/
[8] Same methodology as described for AI mentions below — Using Winter Fuel Payment and derivations as keywords
[9] https://www.ncsl.org/technology-and-communication/artificial-intelligence-2025-legislation
[10] https://www.cov.com/en/news-and-insights/insights/2023/10/us-artificial-intelligence-policy-legislative-and-regulatory-developments
[11] https://www.nafa.org/publication/government-action-in-ai-and-transportation/?utm
[12] https://www.dlapiper.com/en-us/insights/publications/ai-outlook/2025/ten-year-moratorium-on-ai
[13] Wang, Z., Xie, Q., Feng, Y., Ding, Z., Yang, Z. and Xia, R., 2024. Is ChatGPT a Good Sentiment Analyzer?. In First Conference on Language Modeling.
