HUMAIN ALERT

← All seven themes

Understand. Choose. Act.

AI & our future

Recognise the risks. Keep human judgement. Use AI to help people.

Choose an action
A woman and a humanoid robot face each other with curiosity
AI-generated illustration

Start with what you can influence.

AI can make work and learning easier, but it can also produce convincing mistakes, expose private information and help scammers. The three actions below focus on choices you control. Wider questions about jobs, unfair decisions and powerful AI systems also require action by institutions and governments.

Something you can do today.

Choose an action to begin

Saved on this device only. These are your choices, not a community total.

10 minutes

Check before you trust

An urgent voice message or convincing image can be fabricated. A separate check can stop a bad decision.

  1. Choose someone you trust and discuss urgent requests for money.
  2. Agree to call back using an independently saved number.
  3. Do not transfer money or share codes until the request is verified.

5 minutes

Keep private information private

Documents can contain details about you and other people that an AI service does not need.

  1. Use a made-up example where possible.
  2. Remove names, contact details, account numbers and confidential information.
  3. Follow your workplace rules and check the service’s data settings before uploading.

15 minutes

Use AI to help someone learn

AI can explain a topic in different ways, but it can also make mistakes.

  1. Choose a low-stakes topic and ask for a simple explanation.
  2. Check important facts against a reliable original source.
  3. Share the checked explanation and say how AI helped.

Put it into practice

Practise a calm response to an urgent message

Example scenario: a voice message appears to come from a relative: “I need money now; please do not call.” Treat the urgency as a reason to pause. Recognition of a voice is not independent confirmation.

  1. First two minutes: do not pay, open a payment link or share a code. Leave the message and call the person using the number already in your contacts.
  2. If they do not answer: contact another trusted person through a channel you already use. An unanswered call does not verify the request.
  3. Before the next incident: agree a sentence you can both use: “I will always check money requests by calling back. That is our routine, even in an emergency.”

Useful result: both people know the callback routine. Practising it does not prove you can detect every scam.

For a positive use of AI, ask it to create three practice questions about a school topic from a text you are allowed to share. Check each answer against that text, correct errors and let the learner explain one answer in their own words.

FTC explanation of voice-cloning emergency scams ↗. The practice routine above is Humain Alert’s suggested application.

Illustrative exercise by Humain Alert · 13 September 2026. This is not a reported visitor experience.

Watch & understand
Video thumbnail: Geoffrey Hinton, Nobel Prize in Physics 2024: Official interview
Video source: Nobel Prize

Geoffrey Hinton

A Nobel Prize interview exploring AI. This is a risk perspective, not a prediction of what will certainly happen.

6 December 2024

Watch at the original source ↗

External video. The speaker and publisher are not affiliated with Humain Alert.

Optional reading
Book cover: Co-Intelligence: Living and Working with AI
Book source: Penguin Random House

Co-Intelligence — Ethan Mollick

An accessible guide to working and learning with AI. Published in 2024; specific tools and capabilities have since evolved.

Accessible introduction

Explore this book ↗

No purchase is needed to take action. This is not an affiliate link.

Sources and perspective

A familiar voice is not proof of identity. Checking a message through another channel reduces risk but cannot guarantee safety. Never use an AI answer as the only basis for an important health, legal or financial decision.

Action wording: Humain Alert · reviewed 2026-09-09. See our editorial process. Suggest a correction.

Your experience could help someone else.

Tell us what you tried, or suggest a better action.

Share an experience