Burnham, neoliberalism and AI

Following the Labour Party Conference, there is certainly a new mood of optimism that the Party may be returning to its social democratic roots. While socialist ideas may be in the ascendancy, let’s not get too carried away just yet. Major policy change is still to be realised and implemented. We can all applaud Andy Burnham’s idea of reversing neoliberalism by taking back control of privatised essential services, but how will this be done? What about the future?

A new wave of neo-liberal, generative AI (GenAI), based on Large Language Models (LLMs) and hyperscale data centres is sweeping the globe, driven by US Big Tech oligarchs. They are engaged in a reckless, deregulated, free market battle to win the AI race and make super profits. How should the Labour Government deal with the opportunities and potential threats posed by AI and how should Burnham protect the UK, not just from the neoliberal harms of the past, but those that threaten us today?

Many of these harms are outlined in Karen Hao’s book Empire of AI (previously reviewed on Labour Hub). They concern safety as well as other serious risks – environmental damage caused by vast hyperscale data centres, the worldwide theft of data (both personal and professional), the exploitation of low wage workers in the global south to scrape toxic data from LLMs, the threat to jobs and of a financial bubble and a crash arising from rampant AI speculation. Hao makes the point that there are many forms that AI development could take for the benefit of humanity, but the one chosen by AI oligarchs, large-scale GenAI and LLMs, is potentially the most damaging, but also the most profitable.

Vast sums are being spent by US tech oligarchs trying to influence governments and ensure there is minimal public regulation of their activities. This has been influential in shaping policy, particularly in the US, but also the UK. In fact, Labour’s current AI policy is mainly in tune with the ideological free market, economic growth line pushed by Big Tech. UK policy was shaped by Mandelson, McSweeney and Labour Together and consists of a light touch regulatory regime and a welcoming, ‘pro-innovation’ approach. It was sealed politically  in a “world-leading Tech Prosperity Deal” signed by Trump and Starmer, and immediately led to defence, policing and other contracts with the spytech company Palantir. Tom Watson has recently joined their payroll.

This is supposed to be a clever strategy for delivering economic growth, but critics argue that it is speculative, risky and too ‘Big-Tech-friendly’. In its dash for growth, has the Government risked the possibility of UK policy being ‘captured’ by US Big Tech interests? Is encouraging Big Tech to build numerous hyperscale data centres across the country the best way to deliver the kind of long-term, sustainable growth that would benefit society? Does it protect the national interest?

An alternative that Burnham could quickly turn to is the policy actually promised in Labour’s 2024 election manifesto. This pledged a comprehensive AI Bill to “ensure the safe development and use of AI models by introducing binding regulation on the handful of companies developing the most powerful AI models.” The aim would be to develop AI to help drive growth, but also to safeguard the public interest, by providing guardrails to promote safety and avoid the significant social disbenefits of AI. This election pledge was soon abandoned in favour of much weaker arrangements in the form of Liz Kendall’s “Regulating for Growth” Bill. The mantra of the Starmer government at that time was ‘deregulation to support growth’.

 As a result, the statutory oversight of AI is now fragmented across numerous agencies that focus on safety only once AI has been deployed to the businesses and public services using the tech, rather than the US corporations designing the models. As for the threat of Frontier AI models escaping human control, the AI Security Institute (AISI) was set up as a research and evaluation body. However, it does not have regulatory powers, and companies are not legally required to submit their models or follow its recommendations. It is essentially toothless and has been unable to stop models that it has actually flagged up as problematic from being released, 

The lack of effective control has led to urgent calls for change. Parliament’s Joint Committee on Human Rights (JCHR) issued a highly critical report on 14th September warning that the UK’s AI regulation framework is “entirely unfit for purpose”. The JCHR urged Labour to abandon its voluntary, sector-by-sector approach and immediately enact a risk-based AI Bill. The Government has a month to respond to the report and Burnham certainly has a mandate from the electorate if he chooses to follow this advice!

There is also considerable concern among organised workers. The TUC Congress in September unanimously passed a composite motion: “Artificial Intelligence at work – a just transition through worker voice for the benefit of workers and society”. This consolidated several motions from tech, telecom, finance, and creative sector unions (including the CWU, Usdaw, Aegis, and Equity) into a united call for legislative action to insulate workers from AI exploitation. The motion demanded:

A New National AI Plan: to legally place workers’ rights, environmental safety, and public security on an equal footing with national security and economic growth. 

Resist “Big Tech” Lobbying: by legally protecting strategic state infrastructure and datasets (such as public NHS data) from the undue influence, privatisation and data-harvesting efforts of American technology firms.

An Independent AI Regulator: A standalone domestic enforcement watchdog equipped with legal powers to audit corporate code, issue mandatory compliance notices and fine companies that violate workplace laws.

The motion also called on the TUC to campaign for a second Employment Rights Bill introducing: (1) statutory collective bargaining before the introduction of AI, (2) mandatory impact assessments, to reveal hidden software tracking metrics or encoded biases, (3) penalties for companies that exploit automation to reduce jobs and wages by relocating workflows overseas, (4) protection for entry level jobs and apprenticeships from being quietly replaced by AI systems.

Amendments to protect workers across vulnerable sectors were also included, calling on Labour to take action in certain areas:

Creative Industries (Equity & Writers’ Guild): Enact a statutory regime of “personality rights” to legally block AI developers from cloning or creating unauthorised digital replicas of an actor’s face, body, or voice without explicit consent and fair payment. An ‘opt-in’ structure making it illegal for tech labs to scrape intellectual property, journalism, or scripts to train large language models unless authors explicitly grant permission and receive compensation.

Finance Sector (Aegis): Back the “AI Starts With Workers” charter, which aims to regulate automated profiling systems used by major financial institutions to monitor the keystrokes, breaks and compliance targets of office staff.

Live Arts (Musicians Union): support campaigns targeting “Key Comp” where AI technology is deployed by theatre productions to replicate entire orchestras digitally. Demand live musical performances remain legally protected in West End and touring arrangements.

At the Labour Party Conference a similar motion, backed by the trade unions, was voted into the priority list, but wasn’t actually taken. This was due to ‘time constraints’ and the way the Conference Arrangements Committee managed the agenda.

Nevertheless, in his Leader’s speech, Burnham did refer to AI. He didn’t directly address the key issues raised in the motion, or mention the manifesto pledge of an AI Bill, or the call by the TUC and the JCHR for a domestic statutory watchdog. Instead Burnham took AI oversight to a more personal visionary level announcing that “when Britain takes the Presidency of the G20 next year (in Manchester), I will show our new-found confidence as a country, and I will show it to the world by putting AI centre-stage and Britain will then work to develop a new global code to capture its benefits whilst being clear-eyed about its risks.”

Fine words, but some might say a case of distraction from the issues raised by the unions, or just kicking the can down the road. If Burnham does propose a global code at the G20 what would be its substance and is there any realistic chance of it being achieved? 

It’s all very well saying “we led the world’s first industrial revolution from the Midlands and the North of England. So why can’t we lead the next?,” but the reality is that the US and China are in control of global GenAI development at the macro economic level. In terms of investment and infrastructure, US companies are projected to have invested $764 billion by 2026, (China $102 billion), and to have built 5,427 data centres (China 449). In terms of computing power for advanced AI training, the US controls nearly 75% of global capacity. In contrast the EU only has a 30 billion euro dedicated programme to build seven AI “gigafactories” (leading EU states such as Germany and France are also working with US tech to build data centres, as in the UK).

On the very day that Burnham delivered his speech, Trump announced the “White House Accord on Super Intelligence”, a voluntary pact with the leading US companies relying on ‘self-policing’ rather than federal government regulation. Given that China already has its own statutory regulation system, (as does the EU), the idea that a new global code, led by the UK, can be agreed at the G20 is therefore highly unlikely. Moreover, the G20 takes all decisions exclusively by consensus and joint declarations require full agreement.

Before the government can lead anything, it needs to pause and consider the direction it should take on AI development. It should put its own house in order and implement a thorough review of current AI policy to assess whether it is ‘clear-eyed’ about the risks and benefits of Gen AI. In my view it isn’t and needs a radical overhaul. Despite some protections being put in place, for example, online child safety, and in the creative industries, these are far from sufficient. For example, to protect creative workers, the UK has rejected strict laws like the EU, and has chosen to use traditional copyright laws as a safeguard. By rejecting mandatory tech transparency, the UK currently leaves creators more legally vulnerable than their European counterparts. Under UK law, AI firms cannot scrape copyrighted British data, but they are not required to disclose what data they have already taken and British artists bear the expensive legal burden of discovering if an AI company stole their work.

At present it is unclear whether the government will continue with its neoliberal market-driven policy on AI, allowing, indeed encouraging, Big Tech to build hyperscale data centres across the UK. Burnham acknowledges growing public anxiety regarding job disruption and AI risks, but at the same time stresses a pro-business, pro-innovation approach and the idea of regulation only features in his (to be defined) proposed G20 Global Code.

AI Minister Kanishka Narayan also addressed AI regulation, infrastructure, and its threat to jobs at various events and media interviews at Labour Conference. He said nothing was off the table in considering AI Safety and future legislation. However, there was no commitment to any real policy review or change. He spoke of future contingency plans concerning job losses that could dictate “where we might go on some aspects of regulation” and promised “much better transparency when AI and technology is adopted.” 

He also focused heavily on physical tech infrastructure, demanding that the UK “move faster” to build its data centre capacity. He said this was a matter of national autonomy, arguing that the country needs a “sovereign level of compute“ built explicitly with “a sense of control and a sense of British values,” rather than relying entirely on imported American tech frameworks. However, almost none of the large-scale mega data centres currently under construction or in the planning pipeline (around 90) are actually UK-owned. This  massive infrastructure boom is almost entirely funded, owned, and developed by foreign capital, principally US private equity giants and US tech companies. This is what is “moving fast” and it is difficult to understand why UK policy has been so keen to promote this, if it is concerned about UK sovereignty. Does a “sovereign compute plan” actually exist and does the UK have the funds to deliver it? 

The real danger is that the current level of US Big Tech-funded data centre development could actually lead to over- capacity. As for environmental harm, in August, Labour rejected calls from the Green Party to pause construction of major AI infrastructure, which require vast amounts of power and water. The government hit back at the Green’s proposal of a moratorium, except for local-scale datacentres for the local community, saying it “would be a disaster for jobs and national security”.

As for “moving faster” on data centres and maintaining “sovereignty”, this is also the line being pushed by George Osborne in his new role lobbying governments on behalf of OpenAI. George is now busy accusing data centre objectors of “nimbyism”. The fact is that protests against the harmful impact of data centres are escalating across the UK, as are concerns about AI safety. The government will not be able to ignore rising dissent for much longer, with vague arguments about jobs and national security. There are also concerns about jobs and national security in the way the government is rushing to adopt the Silicon Valley agenda.

In his UN address, Burnham said: “The companies building the models cannot simply be left to their own devices. Fundamental decisions on this technology are too important to be left to chance. They must be taken by elected governments who serve working people.” Under Trump’s voluntary pact the neoliberal, free-for-all will continue. The only way the UK can really ensure control and accountability is by introducing its own regulations as suggested in Labour’s manifesto and now called for by Parliament’s JCHR. This means breaking from the Mandelson/Big Tech, ‘capture’ of UK policy and showing that Labour can really serve working people. The left needs to prioritise this as a key demand.

In developing AI, the UK will of course still have to work with US Big Tech and its infrastructure and models. But the government has to set limits and guardrails on how it does so. There should be a pause in what seems like a reckless rush to build data centres, at least until a comprehensive risk/benefits analysis has been undertaken, debated and made public. Different, less harmful and less costly, ways of using AI under democratic and localised control need to be considered.

 Burnham also talks of moving closer to the EU and there is a clear case for quickly  doing so on AI. The EU already has robust regulation and the ability to develop sovereign infrastructure. A partnership and common approach with the EU would also make the UK much stronger in any negotiations with Big Tech in the US and China. 

Frank Hansen is a former Councillor in the London Borough of Brent.

Image:https://www.picpedia.org/chalkboard/a/artificial-intelligence.html License: Creative Commons 3 – CC BY-SA 3.0 Attribution: Alpha Stock Images – http://alphastockimages.com/ Original Author: Nick Youngson – link to – http://www.nyphotographic.com/ Original Image: https://www.picpedia.org/chalkboard/a/artificial-intelligence.html