I use artificial intelligenceI most days for my work, just like half of British workers do today. Not for writing articles like this but for speeding up processes, updating policy, modernising and driving a charity forward in a far more challenging time. But the very same artificial intelligence system I am using to help protect historic buildings and people in need in a charity that has been in existence for over 800 years is increasingly looking more threatening.

The reality is even more serious than just that. For decades, Silicon Valley has sold humanity a seductive promise: that technology will make us wealthier, healthier, safer and ultimately more free. Artificial intelligence, we are repeatedly assured, represents the greatest technological advance since the Industrial Revolution. It will cure diseases, transform education, accelerate scientific discovery and release humanity from the burdens of everyday work.

But a profoundly disturbing alternative future is already emerging. What happens when the same technology that can revolutionise medicine also helps criminals dismantle the digital infrastructure upon which modern civilisation depends?

What happens when a relatively inexperienced hacker, equipped with commercially available artificial intelligence, acquires capabilities that once required teams of highly trained specialists?

And what happens when the world’s most powerful technology companies and the people behind them discover that the systems making them extraordinarily wealthy are simultaneously becoming instruments of geopolitical destabilisation, economic destruction and social collapse? It’s almost as if that is what they want.

These are no longer questions confined to dystopian science fiction. The evidence is accumulating, and remarkably, some of the most alarming warnings are coming from the very people building the technology.

The democratisation of digital destruction

The greatest immediate danger from artificial intelligence may not be machines developing minds of their own as many Hollywood films depict. It may be something considerably more mundane: like ordinary human beings acquiring extraordinarily destructive capabilities.

Cybercrime historically required specialist knowledge. Up until now, developing sophisticated malicious software, identifying vulnerabilities and successfully penetrating secure networks demanded considerable technical expertise.

Not any more. Artificial intelligence is beginning to dismantle those barriers.

In August 2025, AI developer Anthropic published an extraordinary threat intelligence report. It documented criminals exploiting its Claude technology to carry out activities previously beyond their abilities.

One individual, possessing only basic programming skills, reportedly used AI to develop sophisticated ransomware. Another operation targeted at least 17 organisations, including healthcare and emergency service providers, using AI to assist with theft and extortion.

Anthropic’s own conclusion was unequivocal: “AI has lowered the barriers to sophisticated cybercrime.”

Consider the implications to this new reality.  The result is the potential democratisation of destructive power, in which the sophistication of an attack is no longer determined solely by the training or sophistication of its perpetrator.

There are important limits to understand: AI cannot reliably turn every amateur into an elite hacker, and human expertise remains essential for many advanced operations. Nevertheless, the direction of travel is deeply troubling.

Britain’s security services are already sounding the alarm

The UK’s National Cyber Security Centre (NCSC) has increasingly warned that AI will accelerate cyber threats.

Its November 2026 assessment anticipates that Russia, China and other states will routinely exploit AI-assisted cyber capabilities for geopolitical purposes by 2028.

The NCSC also expects AI-powered malware to be used in a ransomware attack against British critical national infrastructure.

These are assessments of future risks, not declarations that such catastrophic attacks have already occurred. Nevertheless, they come from one of Britain’s most authoritative cybersecurity organisations.

And we are already experiencing many examples; airport closures, flight cancellations, banking systems offline, search engines and social media – the list go on and every week we are reading about these so-called ‘glitches’.

But imagine the consequences of coordinated attacks disrupting airports, hospitals, electricity distribution, banking networks, telecommunications or emergency services. Worse, imagine all those happening simultaneously. An example was the recent Microsoft ‘glitch’ that crashed numerous public services and workplaces.

A modern economy relies on thousands of interconnected systems functioning continuously. Disrupt sufficiently important components and the consequences can rapidly spread beyond the original target. Public confidence evaporates and easily turns into panic.

For the enemies of Western democracies, including authoritarian states, extremist movements and organised criminal networks, this represents an unprecedented opportunity.

 

They may no longer need enormous armies or sophisticated conventional weaponry to cause significant disruption. In some circumstances, software, stolen credentials and AI-assisted capabilities could inflict disproportionate economic damage. And crucially, defending an open society will undoubtedly become substantially more expensive than attacking it.

The men building AI are warning humanity

Perhaps the most remarkable aspect of this technological revolution is that its architects have publicly acknowledged the dangers.

In May 2023, an international group of AI researchers and executives signed a warning organised by the Center for AI Safety.

Its central declaration was stark:

“Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.“

 

Signatories included OpenAI chief executive Sam Altman, Anthropic chief executive Dario Amodei and Google DeepMind chief executive Demis Hassabis.

Frankly, I would read that again. The leaders of organisations developing some of the world’s most sophisticated AI systems publicly endorsed a warning placing catastrophic AI risk alongside nuclear war and pandemics.

Geoffrey Hinton, the Nobel Prize-winning computer scientist widely regarded as one of the pioneers of modern AI, has been particularly outspoken.

In an August 2026 interview reported by the US National Conference of State Legislatures, Hinton observed: “A lot of the public are now worried about it. And I think they’re correct to be worried about it.”

He has identified threats including mass unemployment, political manipulation, cybercrime and loss of human control.

Of course, acknowledging danger is not the same as predicting inevitable catastrophe. AI executives also argue that properly governed systems could deliver enormous economic and scientific benefits.

But a fundamental contradiction remains. How can an industry warn humanity about potentially catastrophic consequences while simultaneously racing to make its technology more powerful, more autonomous and more widely available?

From the electricity bill to the political barricades

Meanwhile, the public is beginning to question the physical cost of the AI revolution.

Enormous data centres are consuming increasing quantities of electricity, demanding grid connections, using water for cooling and requiring billions in infrastructure investment. Slough is home to more than 30 fully operational data centres (with around 40 across the wider town area), making it Europe’s largest data centre cluster and the second-largest globally outside of “Data Centre Alley” in Virginia, USA. Slough’s resident’s have clearly had enough.

The International Energy Agency’s updated analysis projects global data-centre electricity consumption rising from approximately 485 terawatt-hours in 2025 to 950 terawatt-hours by 2030. But energy consumption may prove to be only the beginning of the political backlash.

In Slough, local communities have increasingly started to campaign against data-centre expansion. Concerns include energy prices, water consumption, environmental effects and whether promised employment benefits justify the enormous industrial footprint. In America – the backlash is rapidly becoming political as was evident in the recent mid-term elections. In some states voters in numerous communities approved restrictions on data-centre development.

Research cited by the Los Angeles Times in September 2026 suggested that opposition had delayed or blocked approximately $68 billion of data-centre projects during one three-month period.

Imagine public attitudes following a major AI-assisted cyberattack that disrupts national infrastructure or causes widespread financial losses. Citizens might reasonably ask why communities are expected to surrender land, electricity capacity and public resources to companies whose technologies can also empower those attacking them. That would represent a profound shift in public opinion.

AI companies would no longer be seen simply as creators of innovative products. They could increasingly be viewed as suppliers of dangerous capabilities whose social costs are being transferred to everyone else. The public might conclude that the companies promising to enhance the human experience are helping create the conditions for its deterioration.

The implications for the world’s largest technology businesses could be extraordinary.

Microsoft, Alphabet, Amazon, Meta, OpenAI, Anthropic and others operate within an economic ecosystem increasingly dependent upon massive investments in AI infrastructure.

Investors are committing extraordinary sums in anticipation of substantial future profits. Goldman Sachs estimates that in 2026 alone, just over $1 trillion has been spent on AI infrastructure.

A particularly damaging incident associated with a major AI platform could trigger lawsuits, government investigations and consumer boycotts. A sustained collapse in confidence could undermine investment plans, damage suppliers and expose financial institutions to losses.

The companies celebrated for ushering in a technological golden age might discover that they had helped create the crisis that destroyed their valuations. With growing anger at data centres, the FT’s main headline this morning was: Data centre construction stocks lose $4.2tn as boom in AI trade unravels

The greatest danger: a new global power imbalance

There is, however, another deeply uncomfortable possible reality.

AI could concentrate immense power in a small number of companies, while simultaneously enabling relatively weak hostile actors to inflict disproportionate harm. This is a dangerous combination.

As defensive systems improve, attackers will improve their methods. As restrictions tighten, some may seek alternative platforms, stolen models or less regulated systems.

An international technological arms race could develop in which no major power feels able to slow down, even when recognising the collective risks.

Western governments may fear falling behind strategic competitors. Businesses may fear losing market share. Investors may demand faster commercial deployment. Everyone understands the danger. Nobody wants to be the first to stop. That is precisely how a technology with enormous potential benefits can become a source of systemic instability.

The public backlash could become uncontrollable

History demonstrates that public acceptance of technological change is not unconditional.

Industries once regarded as essential to modern progress have faced sweeping restrictions, expensive litigation and profound reputational damage when the wider consequences of their activities became apparent. Meta (Facebook) TikTok, X (formally Twitter) and others know all about massive regulatory fines and settlement, that primarily involved abuse of data privacy, protection of minors, and antitrust violations.

Artificial intelligence could eventually confront a far greater reckoning.

A severe cyber crisis, evidence of widespread political manipulation or large-scale economic disruption could turn existing scepticism into furious public opposition.

Political parties might compete to promise tougher restrictions. Governments could impose emergency controls on advanced systems. Communities might reject further data-centre construction. Consumers could abandon particular platforms.

Medicine, scientific research, cybersecurity and economic productivity could all suffer.

The challenge, therefore, is not to destroy artificial intelligence. It is to prevent artificial intelligence from becoming a force that destroys trust in the institutions, economies and communities upon which civilisation depends.

Voluntary promises and carefully managed corporate publicity, as used in the past, are unlikely to remain sufficient.

The ultimate technological betrayal

The tragedy unfolding could end up being immense. Humanity may be on the verge of developing extraordinary tools capable of solving problems that have resisted generations of scientific endeavour. But the real catastrophe will be that thousands of ordinary people, driven by greed, fanaticism, political hatred or geopolitical ambition, using increasingly intelligent systems they couldn’t have designed to inflict damage that they could never have achieved alone.

And when that realisation becomes widespread, public anger may turn decisively against the industry responsible for distributing those capabilities. The greatest danger for the AI giants is not simply that their technology could escape their control, it is that the consequences could destroy the public’s willingness to tolerate their existence.

My view is that whilst Silicon Valley promised humanity (not just the Western world) an extraordinary future they may well be remembered for giving its enemies the tools to tear it apart.