A minister addresses the unions
First Secretary Louise Haigh will speak on Tuesday. She will address the Trades Union Congress. Her subject is artificial intelligence.
The speech on 15 September is intended to set out the new government's approach. It is a balancing act. Ms Haigh will state that ministers must 'heed the warnings' from technology industry leaders about the serious risks posed by advanced AI systems. This is a direct acknowledgement of recent, and public, concerns. Yet she will also tell the conference that artificial intelligence has enormous potential for the British economy. The government will not ignore the risks. It will not ignore the benefits either. The core of the message is this dual focus, an attempt to reassure listeners that development will be managed, not halted. She is a senior minister. The choice of speaker signals the topic's importance.
Her appearance at the TUC Congress is significant. The audience is composed of union representatives from across the United Kingdom, a group whose primary interest is the protection of jobs and workers' rights. A Labour government addressing this specific body about a technology widely predicted to cause massive labour displacement is a delicate political manoeuvre. For decades, unions have fought automation. AI presents a new form of that challenge. Ms Haigh's presence is an effort to bring the Labour movement with her, arguing that the technology's advantages can be harnessed safely and for the good of the country, a position that requires trust between the party and its historical industrial allies.
The speech is one of the first clear statements from the new administration on a complex and fast moving policy area. As First Secretary, Ms Haigh holds a position of considerable authority within the cabinet. Her words are not just for the union delegates assembled in the hall. They are for the technology companies building the systems, for international partners considering regulation, and for the wider public. This is the government planting a flag. The message seems simple. Caution is necessary. Progress is also necessary. Finding a practical path that satisfies both of these objectives will be the real test.
The warnings came from inside the labs
The minister’s phrase is not original. It is a direct reply. The warnings Louise Haigh promises to heed are specific, and they are recent. They came from a report in The Guardian newspaper which detailed an alarming statement from three researchers working at the artificial intelligence company Anthropic. Their assessment was bleak. They warned that artificial intelligence could lead to the extinction of humanity. It could happen within a decade. This was not a prediction from a science fiction author or a fringe commentator. It was an expression of grave concern from people intimately involved in developing the technology itself.
Anthropic is one of the world’s leading AI development companies. It builds these systems. The company is a direct competitor to firms like Google DeepMind and OpenAI, operating at the frontier of what is known as generative artificial intelligence. Its primary function is to research and construct the very models that now provoke both excitement and fear across global governments. The researchers were not senior executives. They were people on the inside. Their job is to push the capabilities of these systems forward, making the warning all the more powerful because it argues against the potential consequences of their own success. This is what makes the government listen. The builders are raising the alarm.
Such a warning carries unusual political weight. It is not easily dismissed. The technology industry typically speaks with a voice of coordinated optimism, focusing on potential benefits while framing risks as manageable engineering problems that can be solved with more innovation. The Anthropic researchers broke this mould. Their public statement bypassed official company channels, presenting a direct challenge to the sector's narrative of inevitable, safe progress. These are not people opposed to technology. They are its creators. This status lends them a unique authority that politicians find difficult to ignore, especially when the subject is as complex and poorly understood by outsiders as artificial intelligence. Their expertise is the source of their credibility. Their fear is what gets attention.
This specific warning is the immediate context for Ms Haigh’s speech at the TUC Congress. It provides the 'why'. Without the public intervention of the Anthropic researchers, her address might have been a more general discussion of economic opportunity and labour market transition. Instead, she is compelled to address existential risk. The intervention has also amplified calls from within her own party for more decisive action. Labour MPs and peers have been pushing the government to increase cooperation with international partners on a framework for AI regulation. The issue has moved from a theoretical policy debate to an urgent political demand. It is a security matter. The First Secretary’s words are therefore a direct acknowledgement that these fears are being heard at the highest level of the new government, a necessary first step before any concrete policy can be formed.
Labour confronts a difficult balance
This is a difficult balance. The government confronts three competing pressures. Each pulls in a different direction. Louise Haigh’s speech is an attempt to hold the centre, an exercise in navigating a problem where the ideal outcomes for one group represent the worst fears of another. The government wants economic growth. The unions want to protect jobs. The developers want to prevent extinction. Ms Haigh must speak to all three audiences at once. It is an almost impossible task. Her solution is to acknowledge every risk while committing to no single path, a strategy of deliberate ambiguity designed to keep all parties engaged while the government decides what to do next.
The first pressure is economic. It is the most straightforward. The government wants the growth that artificial intelligence promises. AI offers a route to productivity gains that could reshape the British economy, a prospect no new administration can ignore, especially one inheriting fiscal challenges. Ms Haigh spoke of its ‘enormous potential’. This potential is the primary motivation for government engagement. It is the prize. The vision is one of automated efficiency in public services, renewed competitiveness for British industry, and the creation of entirely new high skilled sectors. It is a powerful lure. This economic imperative pushes the government towards a light touch regulatory approach, one that encourages investment and rapid development. Do not stifle the golden goose. That is the message from this corner.
Her audience embodies the second pressure. The unions have different concerns. The Trades Union Congress exists to protect the interests of workers, and its members see AI not as an economic opportunity but as a direct threat to their livelihoods. Their fear is not abstract. It is concrete. It is about jobs lost, skills made obsolete, and wages driven down by automation. They worry about the call centre worker replaced by a chatbot, the graphic designer outcompeted by an image generator, the warehouse operative managed by an algorithm. For a Labour First Secretary, this is an acute political problem. She is addressing the party's industrial heartland. Their support is foundational. Her speech must therefore acknowledge their anxieties and imply a commitment to managing the transition, even as she promotes the technology causing the disruption.
The third pressure is the most unusual. It comes from the creators. The Anthropic researchers’ warning of existential risk injects a wholly different dimension into the debate, shifting it from economics to security. This is not a lobby group demanding subsidies or a union negotiating pay. It is a warning from the engine room. They argue that the technology the government hopes to harness for prosperity could, if left unchecked, lead to catastrophic outcomes. This forces the government to take safety seriously, pulling it towards stricter regulation and international cooperation, the very things that might slow down the economic gains it desires. Ms Haigh’s promise to ‘heed the warnings’ is a direct response to this pressure. It is a recognition that the debate is no longer just about GDP figures and employment statistics. It is about control. The government is therefore caught between the promise of wealth, the fear of unemployment and the threat of oblivion.
Regulation is harder than it looks
Labour MPs call for international action. It is an admission of a difficult truth. National regulation alone cannot work. The problem is global. The code has no passport. A law passed in Westminster does little to constrain a company developing its foundational models in a San Francisco office, training them on data centres in Iceland and deploying the finished product to a user in Swansea through an app hosted on servers in Ireland. The object of regulation, the AI itself, is intangible and can be moved or copied across borders almost instantly, making purely national rules feel like trying to build a dam in one small section of a worldwide flood. It is a borderless technology.
This speed and placelessness make it unlike anything the government has tried to regulate before. Consider the railways. The tracks are fixed in British soil. The trains run on them. The companies have physical headquarters in London or Derby. The safety standards for building a carriage or maintaining a signal box can be written, inspected and enforced by British regulators because the entire system is geographically contained. Legislation can be precise. It works. The same is true for food standards or car manufacturing. The factory exists. The product can be tested at the border. You can stop a dangerous car from being sold. It is much harder to stop a dangerous algorithm from being downloaded.
The legislative process itself is too slow. It takes years. A bill is proposed, debated, sent to committee, amended by the Lords and eventually given Royal Assent. By the time a law designed to manage the risks of one generation of AI is passed, the technology will have moved on through two or three more generations, each one exponentially more powerful and presenting entirely new problems the original drafters could not have imagined. The law would be obsolete on the day it was enacted. This creates a dilemma for ministers. Do nothing and risk being accused of complacency. Act too quickly and risk creating bad law that stifles innovation without actually improving safety.
Therefore, the only plausible path is international cooperation. Yet this has its own problems. The European Union has its AI Act, a vast and complex piece of legislation. The United States has pursued a lighter touch through presidential executive orders. China is building a completely separate, state controlled AI ecosystem with its own rules. Getting these powerful blocs to agree on a single, coherent set of binding safety standards for the most advanced AI is a diplomatic challenge of the highest order. It requires aligning fiercely competing economic and geopolitical interests on a subject that is still not fully understood. That takes time. It takes immense political will. It is the only way that might work. It is also the hardest.
What to watch for next
Louise Haigh's speech is a statement of intent. It is not policy. The real test of the new government's position will be the specific actions it takes over the coming months and years, turning the rhetoric of a TUC Congress address into the hard substance of legislation, spending commitments and international treaties. Attention will now turn to the legislative agenda set out in the King's Speech and the financial priorities revealed in the Autumn Statement. Words are cheap. Action is expensive. These events will provide the first concrete evidence of how seriously ministers are taking the warnings from inside the technology companies. They will show whether this is a core priority or just another problem to be managed.
One key signpost will be the government's approach to domestic regulation. A new, dedicated AI regulator is a distinct possibility. This is an option. Such a body, perhaps modelled on Ofcom for communications or the Financial Conduct Authority for finance, could be staffed with technical experts and given powers to adapt quickly to a fast changing technology without waiting for new acts of parliament. The alternative is to distribute responsibility among existing regulators, giving the Information Commissioner's Office power over data, the Competition and Markets Authority power over monopolies, and the Health and Safety Executive power over workplace impacts. This would trigger a Whitehall battle. It would be a fight for control. Each department would see an opportunity to expand its influence and budget, creating a complex and potentially fragmented system of oversight that powerful technology companies might find easy to navigate and exploit.
Funding for safety research is another critical area. A decision is needed. The government could make a significant financial commitment to create a public sector check on the work being done inside private labs, a direct response to the concerns raised by the Anthropic researchers. Tech companies spend billions of pounds developing their AI systems but only a tiny fraction of that total is spent ensuring those systems are safe. A new national AI safety institute, publicly funded and independent, could be tasked with auditing the most powerful models for dangerous capabilities, from autonomous cyberattacks to bioweapon design, before they are released. This costs money. It would require hundreds of millions of pounds. Such an announcement in the Autumn Statement would be the most definite signal that Haigh's promise to 'heed the warnings' is being taken seriously at the highest levels.
Ultimately, the UK's actions will be measured on the international stage. The ambition is clear. The government wants to position Britain as a global leader in AI safety, building on the foundation of the Bletchley Park summit. The next step is to translate that ambition into tangible diplomatic progress. That is hard. The true challenge involves pushing the United States, the European Union and China towards a shared, binding agreement on safety standards for the most advanced forms of artificial intelligence, a task requiring the alignment of profoundly different economic systems and competing geopolitical goals. This may prove impossible. For now, a speech to the unions has set a direction. The machinery of government must now follow.
Sources. BBC News UK: We must heed warnings of AI tech developers, says UK minister. Guardian UK: UK must heed warnings from AI experts, minister says. Evening Standard: UK will ‘heed the warnings’ about AI risks, senior minister Louise Haigh to say.
Analysis. Drafted with AI assistance from the sources listed above and reviewed by an editor before publication. Jnews links to the organisations it writes about.
