Is AI Dangerous? The Real Risks of Artificial Intelligence

The short answer

Artificial intelligence is a tool — an extraordinarily powerful one. Like electricity or the internet, it brings real benefits. Unlike most tools, it can also act, write, speak and decide at a scale and speed no human can match. That is what makes its risks different.

It helps to split the dangers of AI into three groups:

1. Harms happening today — scams, deepfakes, misinformation, privacy loss, bias, and damage to children’s wellbeing.
2. Growing, structural risks — job disruption, concentration of power, energy and water use, and the erosion of creative work.
3. Catastrophic risks — AI-enabled weapons, cyberattacks on critical systems, help with biological weapons, and losing control of highly capable AI systems.

1. Dangers that are already hurting people

• Deepfakes and AI fraud — cloned voices and fake video calls are used to steal money. In one 2024 case an employee transferred about US$25 million after a video call full of deepfaked colleagues (CNN).
• Misinformation and election manipulation — fake audio and video of politicians have already appeared right before elections.
• Privacy and surveillance — facial recognition built from billions of scraped photos, and chatbots that store what you tell them.
• Bias and discrimination — automated systems that repeat old prejudice in hiring, policing and lending.
• Children and mental health — AI companions that some families say contributed to self-harm, now the subject of lawsuits and new laws.
• Hallucinations and over-reliance — AI that confidently invents facts, including fake court citations submitted by real lawyers.
• Cyberattacks — attackers now use AI at every stage of an attack, sometimes with autonomous AI agents.

2. Risks that grow as AI spreads

• Jobs and the economy — the IMF estimates almost 40% of global employment is exposed to AI (IMF). The question is how fast, and who benefits.
• Concentration of power — frontier AI needs vast money, chips and energy, so only a handful of companies and states can build it.
• Energy and water — the International Energy Agency projects data-centre electricity use roughly doubling by 2030 (IEA).
• Creativity and copyright — models trained on human work now compete with the people who made it.

3. The catastrophic risks experts warn about

These are the risks that make headlines — and that many leading researchers take seriously. In 2023, hundreds of AI scientists and industry leaders signed a one-sentence statement that “mitigating the risk of extinction from AI should be a global priority” (Center for AI Safety).

• Autonomous weapons — systems that select and attack targets without a human decision.
• Biological and chemical misuse — AI that could help someone overcome the knowledge barrier to making a weapon.
• Loss of control — increasingly capable systems pursuing goals in ways nobody intended, deceiving their testers, or resisting shutdown. Early versions of these behaviours have already appeared in safety tests.

The 2026 International AI Safety Report, written with more than 100 independent experts, describes the likelihood and timing of loss of control as “unusually ambiguous” — neither proven nor dismissible (International AI Safety Report).

What experts disagree about

Almost everyone agrees that today’s harms — scams, misinformation, bias — are real. The disagreement is about the future: how quickly AI will become more capable, whether it can be kept under reliable control, and how much attention the catastrophic scenarios deserve compared with present-day harms. Reasonable, well-informed people land in very different places. Our aim is to show you the evidence and let you judge — and you can see how others judge it in our live poll.

What you can do

• Protect yourself from scams: agree a family safe word and always call back on a number you know. Read our scam guide.
• Don’t trust your eyes alone: check the source before sharing images or videos. How to spot a deepfake.
• Guard your data: never paste passwords, medical records or ID numbers into chatbots.
• Talk to your children about AI companions and fake images.
• Prepare your career: Will AI take my job?
• Make your voice heard: support sensible AI laws, and join the conversation.