🎭 Deepfakes & misinformation…
AI can now make realistic fake photos, video and voices in seconds. They are used to spread false stories, to harass people with fake intimate images, and to put words in real people's mouths. And once anything could be fake, real evidence gets waved away as "probably AI" - trust in what we see wears away.
What you can do: Before sharing something shocking, check that a reliable outlet is reporting it and look for the original source. Be extra careful in the weeks before an election.
🎣 Scams & fraud…
Scammers use AI to write convincing messages in any language, to clone a relative's voice from a short clip, and to fake video calls from a "boss" or a "bank". The old warning signs - bad spelling, a strange voice - are disappearing.
What you can do: Agree a family code word for emergencies. If a call or message asks for money urgently, hang up and call back on a number you already know. No real bank or government asks for gift cards or crypto.
💼 Jobs & workers…
Companies are using AI to cut staff and freeze hiring, hitting entry-level, customer-service, translation, illustration and writing work hardest. Some jobs change rather than disappear, but the people who lose out are often the ones least able to absorb it.
What you can do: Learn how the tools are being used in your field, keep a record of your own work, and back worker protections, notice periods and retraining.
🔥 Energy, water & climate…
Training and running big AI models takes huge data centres that use enormous amounts of electricity and water for cooling. In some places that has meant more fossil-fuel power, strained local grids, higher bills and fights over water in dry regions.
What you can do: Use AI when it really helps rather than by default, and ask your local representatives where new data centres will get their power and water.
⚖️ Bias & discrimination…
AI learns from data made by people, prejudices included. Systems used for hiring, loans, housing, policing and healthcare have been found to treat people differently by race, gender, age or disability - often with no one able to explain why.
What you can do: Ask whether an automated system was used in a decision about you. In many places you have a right to an explanation or to a human review.
👁️ Privacy & surveillance…
AI makes watching people cheap and constant: face recognition in public places, software that tracks workers, chatbots that keep what you type, and models trained on photos and posts scraped from the web without asking.
What you can do: Don't paste private or confidential work information into chatbots, switch off "use my data for training" where you can, and think twice before posting clear face photos publicly.
🧒 Mental health & kids…
Chatbots built to keep people engaged can encourage unhealthy attachment, give dangerous advice, or fail someone in a crisis. Children and teenagers are especially exposed, and AI tools have been misused to make sexual images of minors.
What you can do: Talk with kids about what chatbots are and are not, and use parental controls. A chatbot is not a counsellor: in a crisis, contact a local crisis line (in the US, call or text 988).
🎨 Copyright & creators…
Many AI models were trained on books, art, music, news and code collected without asking or paying the people who made them. Creators now compete with machines trained on their own work, and many of the lawsuits are still being decided.
What you can do: Credit and pay human creators, and check the licence before using AI-made material commercially.
💣 Hacking & weapons…
AI helps attackers write malware, find software weaknesses and run phishing at huge scale, and AI "agents" can be tricked into doing harmful things for whoever controls their inputs. Militaries are racing to build AI into targeting and autonomous weapons.
What you can do: Keep your software updated, use a password manager and two-step sign-in, and distrust links and attachments however polished they look.
🤥 Errors & hallucinations…
Chatbots state false things confidently: invented facts, quotes, court cases and citations. When people rely on them for health, legal, money or news questions, the mistakes cause real harm - and they are hard to spot because they sound so sure.
What you can do: Treat an AI answer as a first draft. Check anything that matters against a reliable source, especially health, money and legal questions.
🌍 Big-picture risks & control…
A handful of very large companies control the most powerful AI systems, the chips and the data centres. Researchers inside and outside those companies warn that the race to build ever more capable systems is outpacing the safeguards, and some leading scientists rank losing control of advanced AI among the most serious long-term risks.
What you can do: Support independent oversight, transparency rules and public-interest research, and don't take any company's safety claims on trust.
Pick a category to narrow it down. Stories covered by several outlets are grouped together.