Should we fear AI?

Talk to business owners in Hanoi or Jakarta about artificial intelligence and you will often hear excitement. Have the same conversation in Amsterdam, Paris or London and the mood is more cautious. The surveys confirm the impression. In Stanford University's AI Index, which uses Ipsos data from 2024, 83 per cent of people in China and 80 per cent in Indonesia said AI products have more benefits than drawbacks. In the Netherlands the figure was 36 per cent, in France 41 per cent, in the United Kingdom 46 per cent and in Germany 47 per cent.

The divide is not simply East and West. Pew Research Center found in 2026 that concern about AI is especially widespread in high-income countries, and Japan and South Korea are cautious too. European optimism is also rising: in France, the share seeing more benefits than drawbacks went from 31 per cent in 2022 to 41 per cent in 2024. Caution is not a weakness. It can be a strength, if it leads to good decisions rather than paralysis.

The film version and the real questions

Cinema has given us killer robots and machines that decide humanity is the problem. These stories are memorable, but they can distract from the questions scientists actually study. The International AI Safety Report, prepared by more than 100 experts under the chairmanship of the computer scientist Yoshua Bengio and backed by over 30 countries, sorts the risks into three groups: malicious use by people, malfunctions such as errors and bias, and wider effects on society.

Could AI escape our control?

On the scenario people fear most, the report is clear about today: current systems lack the capabilities to pose such a risk. About the future, experts genuinely disagree. In a large survey of 2,778 AI researchers published in 2024, the median estimate of an extremely bad outcome, such as human extinction or severe loss of control, was 5 per cent, with very wide differences between individuals. Some leading scientists, including Geoffrey Hinton and Yoshua Bengio, have signed statements that such risks should be a global priority. Others, such as Yann LeCun, call the takeover scenario far-fetched, and the Princeton researchers Arvind Narayanan and Sayash Kapoor argue that AI is best seen as a normal technology that we can and should keep under control.

What do the AI developers themselves say? They are surprisingly frank. Anthropic has written that nobody yet knows how to train systems to behave well in every situation, and OpenAI has warned that a misaligned superintelligence, or an autocratic regime with a decisive lead in AI, could cause grievous harm. Ask a modern AI assistant whether AI could become dangerous and it will usually give a similar answer: a serious question on which experts disagree, with today's systems still tools under human direction.

Weapons: partly real already

Autonomy in weapons is not science fiction. In Ukraine, according to a 2025 report by the Center for Strategic and International Studies, AI helps drones on the final stretch to a target that a human has selected, raising hit rates considerably. Fully autonomous killing without a human decision is not the documented norm, but the direction of travel is clear.

The world is trying to respond. In 2024, 166 countries voted for a UN General Assembly resolution on lethal autonomous weapons, with only three against. The UN Secretary-General and the President of the International Committee of the Red Cross called for a binding treaty by 2026, saying that the autonomous targeting of humans by machines is a moral line that must not be crossed. Talks in Geneva ended in September 2026 without a treaty; the next opportunity is the review conference in November 2026. Defence is excluded from the EU AI Act, so this is a matter for governments and international law, and one where public opinion matters.

The risks that are here now

For most of us, the immediate dangers are more ordinary. In early 2024 an employee of the engineering firm Arup in Hong Kong transferred about 25 million US dollars after a video call in which the chief financial officer and colleagues were all deepfakes. The International AI Safety Report notes that criminal groups already use AI in cyberattacks. Misinformation, errors from trusting AI output without checking, and the concentration of power in a few hands are the concerns researchers rate highest. In the same 2024 survey, 86 per cent of researchers called deepfakes and misinformation a substantial or extreme concern.

Protection is possible

Europe is not standing still. Since February 2025 the EU AI Act has banned practices such as social scoring and manipulative AI, and providers of general-purpose AI now have obligations on transparency and safety. The Council of Europe's Framework Convention is the first binding international treaty on AI, and governments have set up AI safety and security institutes. None of this is perfect, but it is a foundation.

Companies can protect themselves too, mostly with sensible habits. Confirm any payment request or change of bank details by calling back on a known number. Train staff to recognise deepfakes and voice cloning. Keep confidential data out of public AI tools. Keep a human check on important AI output. And make sure people understand the tools they use.

Worry now, or prevent now?

Our view is that fear is the wrong guide, and so is indifference. The serious scenarios are not happening today, and scientists agree that nothing about the future of AI is inevitable. That is exactly why the present matters. The direction AI takes depends on choices made now: by governments on weapons and oversight, by developers on safety, and by every company on how it uses these tools. Calm attention and sensible prevention are better than either panic or complacency.

Published 1 October 2026 · © 2026 Habsburg Digital Ltd