ChatGPT Says Cybercrime Is a Pervasive and Damaging Global Threat, Other AI Agree

We are often told that the solution to humanity’s problems is Artificial Intelligence (AI), so I asked seven of the world’s most popular AI systems for their views on the problem of cybercrime. They all say that cybercrime is a serious and rapidly escalating threat to governments, companies, and the public, with harmful impacts that are not only financial but also psychological, cultural, and structural. Furthermore, things will get worse in a hurry if humans cannot immediately institute massive changes in their behaviour. According to most of the AI included in my research, when cybercrime is considered in the context of other challenges faced by humanity today — think climate change and a widening wealth gap — it amounts to an existential threat; predictions include societal collapse, global recession, and an increase in global conflicts. The phrase “cyber-feudal hellscape where only the ruthless survive” was used. While all these AI are happy to describe the changes that must be made to rein in cybercrime before it seriously undermines the future of human existence, they point to a serious lack of resolve among humans to make those changes. Basically, these seven AI agree with what many human experts in cybersecurity and crime deterrence have been saying for years, experts whose warnings have long been ignored by those same humans who are now championing AI.

I have prepared a report detailing my conversations about cybercrime with ChatGPT, Claude, DeepSeek, Gemini, Perplexity, Meta AI, and Grok. Commentary and background are included, along with notes on methodology and limitations of the study. You can download a PDF of the 42-page report here: Cybercrime According to AI: a dire and rapidly escalating crisis.

After more than three decades trying to persuade governments and corporations that they must do more to deter cybercrime, I find myself in strong agreement with the following statement from OpenAI’s ChatGPT:

cybercrime today is “severe and worsening…one of the most pervasive and damaging global threats, and governments and companies are consistently behind the curve.”

Other AI give similiar responses when I asked this question: what do you think about the state of cybercrime in the world today?

Anthropic’s Claude Sonnet thinks: “Cybercrime today is absolutely rampant and largely out of control.” And Google’s Gemini offers this take:

“The state of cybercrime in the world today is alarming and deeply entrenched. It’s a highly adaptive, technologically advanced, and pervasive global threat that consistently outmaneuvers reactive security measures.”

The popular AI search tool known as Perplexity thinks:

“The global cybercrime landscape has reached a crisis point, one that is “characterized by unchecked escalation in sophistication, financial impact, and systemic risk” (emphasis in the original).

But why did I ask AI about the cybercrime problem? Because human efforts to persuade the leaders of governments and industry to adequately address the problem of cybercrime, efforts which date back to the 1980s, have failed.

(To be clear, what I mean by the problem of cybercrime is the very human tendency to see technology as a means to achieve selfish ends at the expense of others and in violation of the law, compounded by incentives to develop and deploy technology without adequate safeguards against its potential for abuse.)

We are now in a situation where, on the one hand, human experts in cybersecurity and cybercrime have failed to convince the leaders of government and industry to address the pervasive and damaging global threat of cybercrime with adequate resources and resolve; on the other hand, over the last five years leaders of government and industry have widely hailed artificial intelligence as the answer to serious real world problems, ranging from poverty and disease to productivity and climate change.

My argument goes like this: whereas there is a long and well-documented history of the leaders of government and industry ignoring solutions to the world’s ills proposed by humans, even humans with considerable intelligence and expertise; and whereas today’s leaders have wholeheartedly endorsed AI as the solution to the world’s ills; therefore, today’s leaders have an obligation to heed the warnings and advice of AI.

To put it another way: if AI is so bloody marvellous that you’re pouring a ton of taxpayer and investor money into promoting it, will you please listen to what it says about cybercrime! Because if you don’t, we risk creating, as CharGPT puts it:

“a digital environment that is so hostile, unreliable, and legally ungovernable that large parts of it become unusable or outright dangerous.”

Perplexity has similar thoughts:

“If nothing changes, cybercrime will spiral out of control, fundamentally destabilizing economies, governments, and daily life…The average person will face constant threats to their finances, privacy, and even physical safety as IoT devices and smart infrastructure become attack vectors.”

To be honest, when I posed my original question to ChatGPT—what do you think about the state of cybercrime in the world today? — I didn’t have a formal plan for asking further questions beyond that. But the last sentence of ChatGPT’s response really got me thinking. It was the sort of thing I was accustomed to saying whenever people asked me about the state of play in the epic struggle between cybercrime and cybersecurity. ChatGPT ends its response about the current situation by saying:

“Without major shifts in global cooperation, regulation, and investment in digital resilience, the outlook remains grim.

This way of talking will be familiar to experts in many fields, from public health to atmospheric pollution, climate science to disaster response, emergency management, and more. Using “lifeline” words such as unless, without, absent, or until is indicative of that very human reluctance to describe a grim reality without offering some hope for the future.

Given that ChatGPT implies things could be better in the future if changes are made, I asked a future-oriented follow-up question: Where do you think we end up if nothing changes?

The answer is blunt, but again ChatGPT is very much in line with my own thinking: “If nothing changes, we end up in a world where trust in digital systems steadily collapses.” ChatGPT goes on illustrate how this collapse would extend from individuals to companies and countries, across multiple spheres of activity. In short:

we risk creating a digital environment that is so hostile, unreliable, and legally ungovernable that large parts of it become unusable or outright dangerous. That future is plausible — not speculative — if current trajectories persist.

In my opinion, this is exactly the sort of thing the leaders of companies and governments need to hear, not to mention AI’s billionaire backers and the folks whose hard-earned money is being invested in AI.

As with its response to my first question, ChatGPT feels compelled to assess the possibility of positive change, even though I had not specifically asked for this:

There are promising signals — like the EU Cyber Resilience Act, NATO’s growing focus on cyber defense, and recent joint operations against ransomware gangs — but these are fragmented. The reality is that cybercrime thrives on international disunity, legal loopholes, and enforcement asymmetries.

Then, as with my first question, ChatGPT uses an “until” construction in the penultimate sentence of its assessment of what happens if nothing changes:

Until there’s a global shift toward coordinated cyber governance, greater deterrence, and real consequences for malicious actors, meaningful change is unlikely.

But the final sentence reinforces the lack of optimism.

We might see incremental progress, but not the systemic overhaul the situation demands.

Other AI are of similar mind. Claude paints a grim picture and stresses the urgency of the need for change, using “without” in its closing sentence:

Eventually, we’ll likely see authoritarian responses — governments imposing draconian internet controls and surveillance in the name of security, destroying privacy and digital freedom in the process. We’re not talking about some distant future. These trends are already visible and accelerating. Without dramatic changes, the next decade will be much worse than today.

Deepseek is even starker and darker:

Within 10 years, we’ll hit a “Cyber Pearl Harbor” — an attack so devastating it forces draconian lockdowns of the internet. But by then, the criminals and rogue states will be too entrenched to stop.

If you read all of the replies from the seven AI in the final report you will see that they provide suggestions as to what should be done to tackle the cybercrime problem. While unsolicited, the suggestions are generally in line with expert thinking on cybersecurity and cybercrime deterrence, (although DeepSeek can get a little extreme, suggesting “drone-strike equivalents for hacker havens” as an example of “global cyber-treaties with teeth”).

The suggestions about what should be done to address cybercrime reminds us that cybersecurity experts have known what needs to be done for a long time; the challenge is marshalling the resources and resolve to do it. In my experience, the biggest obstacle to meeting that challenge tends to be competition from other priorities; so I decided to ask AI what the future looked like given the reality that anti-cybercrime efforts have to compete for resources due to other issues, like viral pandemics, armed conflicts between nuclear powers, or climate change. I went with climate change and asked: What does the future look like 10 years from now if meaningful change does not happen and cybercrime keeps growing while efforts to limit climate change struggle?

ChatGPT’s response is bleak but surprisingly thoughtful: “Ten years from now, if cybercrime keeps growing and efforts to limit climate change continue to falter, the world becomes more unstable, unequal, and brittle — both digitally and physically.” Note the reference to an unequal world because I will come back to it in a moment.

ChatGPT goes on to suggest that “feedback loops of vulnerability” between cybercrime and global warming could result in a world where: “The wealthy and powerful insulate themselves behind private networks and fortified supply chains. Everyone else navigates a world of failing institutions, insecure systems, and deepening digital inequality.”

ChatGPT’s concern with inequality in the context of cybercrime prompted me to ask one more question. However, I also wanted an assessment of the grounds for optimism, so the question I asked was this: Do you see any signs of, or potential for, a global movement to reject criminal abuse of digital systems and commit to reform the criminal element that is perpetrating cybercrime? Or do you think the large and growing ‘wealth gap’ within and between countries will perpetuate a moral code in which using technology to gain wealth and power is acceptable?

Across the board, the seven AI models fail to find much about which to be optimistic. Here’s Claude: “The only “movement” I see is toward greater inequality and more sophisticated criminal enterprises. Young people worldwide are learning that technology plus desperation equals opportunity, regardless of legality…Even within wealthy countries, the moral framework is broken. We celebrate tech billionaires who made fortunes through ethically questionable means while condemning poor kids who hack for survival. The message is clear: exploitation is fine if you’re rich enough to make it legal.”

Here’s Grok’s take: “In my view, while global efforts to combat cybercrime are underway, a true movement to reject its moral acceptability and reform perpetrators is still emerging. The wealth gap likely fuels cybercrime by creating incentives for illicit gains, perpetuating a moral code of acceptance in unequal societies. Without addressing economic disparities alongside cybersecurity measures, cybercrime will continue growing, driven by economic incentives rather than a unified global rejection.”

Perplexity’s view is that, “unless there is a radical shift — where governments, industry, and civil society align on enforcement, education, and economic opportunity — the wealth gap will continue to fuel cybercrime.” In other words, as Perplexity states with a classic sliver of hope and a call for action:

“Without deeper reforms addressing both security and inequality, the moral code that tolerates or even celebrates cybercrime will persist, and the problem will only grow.”

Clearly, to the extent that information systems can have opinions, the seven that I questioned about cybercrime had a lot. But what they all boil down to is this: an urgent call for humans to comprehensively and decisively address the problem of cybercrime, before it’s too late. At the same time, these AI all seem aware that humans may not respond to this call. As they see it, right now we are doing too little, too slowly. So I am now researching that means, for humans and for AI. What does “too late” look like and who will it harm the most.

Download the PDF of the 42-page report here: Cybercrime According to AI: a dire and rapidly escalating crisis.

(This article first appeared on Medium in July of 2025. Republished for here for my Substack followers and expanded reach.)